Form Design Patterns

In memory of my
beautiful and amazing

mum, Libby Silver.
Published 2018 by Smashing Media AG, Freiburg, Germany.
All rights reserved.
ISBN: 978-3-945749-73-9
Cover design: Espen Brunborg

Copyediting: Owen Gregory
Interior layout: Markus Seyfferth
eBook production: Cosima Mielke
Typefaces: Elena by Nicole Dotin, Mija by Miguel Hernández.
Form Design Patterns was written by Adam Silver

and reviewed by Heydon Pickering.
Please send errors to: errata@smashingmagazine.com

Table Of Contents
1 A Registration Form 18
2 A Checkout Form 68
3 A Flight Booking Form 123
4 A Login Form 200
5 An Inbox 220
6 A Search Form 262
7 A Filter Form 278
8 An Upload Form 314
9 An Expense Form 350
10 A Really Long and Complicated Form 372

About The Author
A
dam Silver is an interaction designer with over 15
years experience working on the web for a range
of companies including Tesco, BBC, Just Eat,
Financial Times, the Department for Work and Pensions
and many others.
He’s particularly interested in inclusive design and design

systems and writes about this on his blog and popular
design publications such as A List Apart. This isn’t his first
book either: he previously wrote ‘Maintainable CSS’ which
is about crafting maintainable interfaces with CSS.
About The Reviewer

Heydon Pickering is a freelance web accessibility consult-
ant, interface designer and writer living in the UK. He
is an author and editor for Smashing Magazine, and he
also works with leading web accessibility specialists, The
Paciello Group, focusing on inclusive design systems.
Heydon has written multiple books on the subject of

accessibility and inclusive design, including “Apps For All,”
“Inclusive Design Patterns,” and “Inclusive Components.”
Acknowledgements
I’d like to thank a number of people who helped me write
this book:
Graham Veal for helping to set up the foundations of the

design system that accompanies the book. His Node.js and
Heroku expertise saved me many hours of pain. John Oates
for reviewing some of my early drafts, which stopped some
of my bad writing habits early in their tracks. Steven Proctor
for reviewing the section on error message design. If you
want to know how to write a good error message, Steven is
your man. Mark Jenkins, friend and unofficial mentor, for
encouraging me to start writing. Owen Gregory for editing
my words to be simpler, plainer, more coherent, and consist-
ent. Espen Brunborg for the book cover that complements the
content and message of the book so well. Markus Seyfferth
for letting me write this for Smashing and producing such a
beautifully crafted book. Heydon Pickering for many things:
championing the book to Smashing; inspiring the book’s
approach (by problem, not principle); being my technical
editor; and if that weren’t enough, for writing the brilliant
foreword. This book would simply not exist without him.
Finally, my wife Jen, for her patience in letting me spend so

much time writing the book while she looked after our two
little ones.
Foreword
E
very so often, someone will point out that I use
blackish text on whitish backgrounds for almost all
my page layouts. And the only comeback I can think
of is that the same approach has worked for hundreds of
billions of publications over the course of hundreds of years.
You know, that old chestnut.
Making a habit of flouting convention will garner you

attention, spark controversy and earnest debate — even
earn you awards. But it will also confound and alienate your
readers and users — the people your work is really meant
for. That is abject failure.
Paradoxically, in a world saturated with rule breaking and

reinvention, a reverence for the straightforward, familiar,
and simple becomes radical. And it’s a welcome revolution,
because interfaces that are obvious are also inclusive. It’s
not a bad thing to be on the nose.
Let me give you an example. Imagine my excitement when

reading this book, to find Adam recommending that form
labels should appear above their respective inputs. Not
off to the side at an angle, not inside the input where the
actual user input should go, and certainly not as some
absurd animated combination of different positions and
orientations at different times.
That’s actually radical, and really refreshing to read. Because

most designers will do anything but the expected. Then I
have to tell them off on behalf of the users they’re forcing to
decipher their interface. Nobody has time for that.
Don’t get me wrong: I’m not saying there’s nothing new in

this book. I learned plenty. It’s just that my reaction was
never “OK, I guess that’s one way of doing it LOL,” and
always “Damn, that’s it — I should have been doing this all
along.” And it turns out that when you combine standard
elements and simple concepts, even daunting components
like the airplane seat chooser can be accessible, logical, and
lightweight.
To me, this book is about simple solutions to would-be

complex problems. As such, it’s not just about forms. But if
you can make forms easy and pleasurable to use (forms!),
then most everything else will be a cinch.
— Heydon Pickering
viii Introduction
Introduction
I remember my first foray into forms. At the turn of the cen-
tury, web design was one of the modules on the information
communication and technology course I took at sixth form
college. My learning mostly consisted of cutting and past-
ing snippets of HTML, CSS, and Javascript. Yes, I came from
the view-source school of web design and development.
My obsession with forms started when — like with any

other HTML element — I tried to cut and paste it. Despite
rendering OK, when I submitted it nothing happened. Fast
forward seventeen years and here I am writing a book about
form design patterns.
Why Forms?
Every meaningful interaction that happens on the web is
achieved by a form of some sort. Without forms, the web
merely becomes a passive experience — just a way to con-
sume content.
Forms allow users to create, update and delete things.

Whether it’s communicating through email, buying a prod-
uct, online banking, or working on a fully-fledged adminis-
trative digital service, forms are always front and center.
At first glance, forms are rather easy to grasp. In less than an
hour, you’ll have text boxes, radio buttons and select boxes
ix
on the page. But their low barrier to entry turns them into
what Heydon Pickering refers to as a “10,000-volt electro-
magnet for attracting usability problems.”1
This is a big part of why I’m writing this book. Typically,

these usability problems come up again and again.
Why Patterns?
Design patterns serve as guidance and solutions to people
solving similar problems over and over. The reason for
design patterns is twofold.
First, instead of solving the same problem from scratch

every time, we can instead use previously designed, avail-
able, recognized, and well-researched solutions. This saves
a lot of time. And we can use that time to solve newer and
perhaps bigger problems.
Second, by solving the same problem in the same way,

users have a consistent and more coherent experience. The
service, app, or whatever it is, becomes familiar. Familiar
interfaces require less effort to operate. Think about it: every
time you encounter a door, you just know that it can be
opened, closed, and sometimes locked.
1 http://smashed.by/idp
x Introduction
Using design patterns for digital experiences or, more

specifically, forms, makes sense too. By the end of the book,
you’ll have many patterns you can use in your own interface
immediately.
Why These Forms?

I first based this book on 50 principles. Originally, each prin-
ciple would become a short chapter. So there was a chapter
called “Always Use a Label” and another called “Placeholders
Are Problematic.”
There are a few problems with this approach to design.

First, rules can be broken — occasionally. Second, evaluating
problems by principle is constraining. Many go together:
when talking about screen readers, for example, it often
makes sense to discuss keyboard users. And sometimes you
have to make trade-offs.
Instead of centering the book on principles, I decided to

revolve it around real problems. That way we can solve
them as we do at work. The result is ten specific problems
to solve, each represented as a chapter. The chapters are
specific, but most of the patterns are reusable and trans-
ferable to many other forms you might be designing. After
all, a pattern should be unique, but reusable across projects
and organizations.
xi
Here’s a chapter rundown.
1. A REGISTRATION FORM
We’ll start with a basic registration form and take a look

at the foundational qualities of a well-designed form and
how to think about them. By applying something called a
question protocol, we’ll look at how to reduce friction without
even touching the interface. Then we’ll look at some crucial
patterns, including validation, that we’ll want to use for
every form.
2. A CHECKOUT FORM
The one thing per page design pattern is a cornerstone of cre-

ating well-designed forms. We’ll look at why that is before
applying it to a checkout flow. After that, we’ll consider flow
and order with a view to breaking down each step of the
checkout flow. Then we’ll look at several input types and
how they affect the user experience on mobile and desktop
browsers, all the while looking at ways to help both first-
time and returning customers order quickly and simply.
3. A FLIGHT BOOKING FORM
We’ll dive into the world of progressively enhanced, custom

form components using ARIA. We’ll do this by exploring
the best way to let users select destinations, pick dates, add
passengers, and choose seats. We’ll analyze native form con-
xii Introduction
trols at length, and look at breaking away from convention

when it becomes necessary.
4. A LOGIN FORM
We’ll look at the ubiquitous login form. Despite its simple

appearance, there’s a bunch of usability failures that so
many sites suffer from. Social media login hasn’t necessarily
helped matters so we’ll cover that too.
5. AN INBOX
We’ll design ways to manage and action email in bulk, our

first look at administrative interfaces. As such, this comes
with its own set of challenges and patterns, including a
responsive ARIA-described action menu, multiple selection,
and same-page messaging.
6. A SEARCH FORM
We’ll create a responsive search form that is readily availa-

ble to users on all pages, and we’ll also consider the impor-
tance of the search mechanism that powers it. Together,
they can make search discoverable, simple, and useful.
7. A FILTER FORM
Users often need to filter a large set of unwieldy search

results. Without a well-designed filter, users are bound to
xiii
give up. Filters pose a number of interesting and unique

design problems that may force us to challenge best practice
to give users a better experience.
8. AN UPLOAD FORM
Many services, like photo sharing, messaging, and many

back-office applications, let users upload images and doc-
uments. We’ll study the file input and how we can use it to
upload multiple files at once. Then we’ll look at the intri-
cacies of a drag-and-drop, Ajax-enhanced interface that is
inclusive of keyboard and screen reader users.
9. AN EXPENSE FORM
We’ll investigate the special problem of needing to create

and add lots of expenses (or anything else) into a system.
This is really an excuse to cover the add another pattern,
which is often useful in administrative interfaces.
10. A REALLY LONG AND COMPLICATED FORM
Some forms are very long and take hours to complete. We’ll
look at some of the patterns we can use to make long forms
easier to manage.
xiv Introduction
What About Principles?

While I’ve moved away from principle-oriented chapters,
there is still an important place for principles in this book.
Without principles, it’s hard to know whether what we’ve
designed is objectively good.
But where should our principles come from? We can either

steal other people’s, or we can define our own. But before we
get to that, let’s see how we get to certain principles in the
first place.
Our principles normally stem from a belief system. We

believe something should be a certain way, typically for
good reason. A banal example, perhaps, would be showing
up on time for meetings: being late (at least, deliberately)
reveals a lack of respect for the other attendees, and the
meeting's purpose. Without a good reason for our belief
system, principles crumble under scrutiny.
This book is about designing forms for the web. It would be

remiss of me, then, to ignore the essence of the web itself.
The power of the web is one of reach and accessibility. Any-
one with a browser and an internet connection gets to use
it. The principles in this book need to align with this notion
— to uphold its inherent qualities.
xv
Frank Chimero talks about this at length in “The Web’s

Grain,” one of my favorite articles on design.2 The main point
of the article encourages us not to aim to tackle complexity,
but do our very best to avoid it in the first place, mostly by
“going with the grain” and embracing the web’s constraints.
It turns out that not only is this the easiest and cheapest
way to design something, but also that users have a better
time operating these simpler interfaces in the end. It’s the
content and functionality users want anyway.
Whatever we build, in the end, is about users. I don’t want

to leave a single person behind if I can help it. The web is for
everyone. I can’t think of a better set of principles than the
inclusive design principles from the Paciello Group.3
These principles are about good design — and good design

is inclusive.
1. Provide a comparable experience. Ensure your inter-

face provides a comparable experience for all so people
can accomplish tasks in a way that suits their needs
without undermining the quality of the content.
2 http://smashed.by/websgrain
3 http://smashed.by/idprinciples
xvi Introduction
2. Consider situation. People use your interface in

different situations. Make sure your interface deliv-
ers a valuable experience to people regardless of their
circumstances.
3. Be consistent. Use familiar conventions and apply

them consistently.
4. Give control. Ensure people are in control. People

should be able to access and interact with content in
their preferred way.
5. Offer choice. Consider providing different ways for

people to complete tasks, especially those that are
complex or non-standard.
6. Prioritize content. Help users focus on core tasks,

features, and information by prioritising them within
the content and layout.
7. Add value. Consider the value of features and how

they improve the experience for different users.
We’ll refer back to these principles throughout the book,
pointing out where something works or not. These princi-
ples should, at least indirectly, hold us to account through-
out the design process.
By looking at common form patterns through the lens of

inclusivity, this book will help you learn how to apply and
reuse conventions that help users complete the task, regard-
less of how they choose or need to use your service.
18 Chapter 1
A Registration Form
L
et’s start with a registration form. Most companies
want long-term relationships with their users. To
do that they need users to sign up. And to do that,
they need to give users value in return. Nobody wants to
sign up to your service — they just want to access what-
ever it is you offer, or the promise of a faster experience
next time they visit.
Despite the registration form’s basic appearance, there are

many things to consider: the primitive elements that make
up a form (labels, buttons, and inputs), ways to reduce
effort (even on small forms like this), all the way through
to form validation.
In choosing such a simple form, we can zoom in on the

foundational qualities found in well-designed forms.
How It Might Look

The form is made up of four fields and a submit button.
Each field is made up of a control (the input) and its associ-
ated label.
A Registration Form 19
A registration form with

four fields: first name, last
name, email address, and
password.
Here’s the HTML:
<form>
<label for="firstName">First name</label>
<input type="text" id="firstName" name="firstName">
<label for="lastName">Last name</label>
<input type="text" id="lastName" name="lastName">
<label for="email">Email address</label>
<input type="email" id="email" name="email">
<label for="password">Create password</label>
<input type="password" id="password" name="password"
placeholder="Must be at least 8 characters">
<input type="submit" value="Register">
</form>
20 Chapter 1
Labels are where our discussion begins.
Labels
In Accessibility For Everyone, Laura Kalbag sets out four broad
parameters that improve the user experience for everyone:1
• Visual: make it easy to see.

• Auditory: make it easy to hear.
• Motor: make it easy to interact with.
• Cognitive: make it easy to understand.
By looking at labels from each of these standpoints, we can

see just how important labels are. Sighted users can read
them, visually-impaired users can hear them by using a
screen reader, and motor-impaired users can more easily set
focus to the field thanks to the larger hit area. That’s because
clicking a label sets focus to the associated form element.
The label
increases the hit
area of the field.
1 http://smashed.by/a11y4all
For these reasons, every control that accepts input should

have an auxiliary <label>. Submit buttons don’t accept
input, so they don’t need an auxiliary label — the value
attribute, which renders the text inside the button, acts as
the accessible label.
To connect an input to a label, the input’s id and label’s for

attribute should match and be unique to the page. In the
case of the email field, the value is “email”:
<label for="email">Email address</label>

<input id="email">
Failing to include a label means ignoring the needs of many

users, including those with physical and cognitive impair-
ments. By focusing on the recognized barriers to people
with disabilities, we can make our forms easier and more
robust for everyone.
For example, a larger hit area is crucial for motor-impaired

users, but is easier to hit for those without impairments too.
22 Chapter 1
Placeholders
The placeholder attribute is intended to store a hint. It
gives users extra guidance when filling out a field — par-
ticularly useful for fields that have complex rules such as a
password field.
As placeholder text is not a real value, it’s grayed out so that

it can be differentiated from user-entered values.
The placeholder’s low-

contrast, gray text is hard
to read.
Unlike labels, hints are optional and shouldn’t be used as a

matter of course. Just because the placeholder attribute
exists doesn’t mean we have to use it. You don’t need a place-
holder of “Enter your first name” when the label is “First
name” — that’s needless duplication.
The label and placeholder

text have similar content,
making the placeholder
unnecessary.
Placeholders are appealing because of their minimal,

space-saving aesthetic. This is because placeholder text is
placed inside the field. But this is a problematic way to give
users a hint.
First, they disappear when the user types. Disappearing text

is hard to remember, which can cause errors if, for example,
the user forgets to satisfy one of the password rules. Users
often mistake placeholder text for a value, causing the field
to be skipped, which again would cause errors later on.2
Gray-on-white text lacks sufficient contrast, making it gen-
erally hard-to-read.3 And to top it off, some browsers don’t
support placeholders, some screen readers don’t announce
them, and long hint text may get cut off.
The placeholder text is cut off

as it’s wider than the text box.
That’s a lot of problems for what is essentially just text. All

content, especially a form hint, shouldn’t be considered as
nice to have. So instead of using placeholders, it’s better to
position hint text above the control like this:
Hint text placed above the text

box instead of placeholder text
inside it.
2 http://smashed.by/nohints
3 http://smashed.by/unreadableweb
24 Chapter 1
<div class="field">
<label for="password">
<span class="field-label">Password</span>
<span class="field-hint">Must contain 8+ characters
with at least 1 number and 1 uppercase letter.</span>
</label>
<input type="password" id="password" name="password">
</div>
The hint is placed within the label and inside a <span> so it

can be styled differently. By placing it inside the label it will be
read out by screen readers, and further enlarges the hit area.
As with most things in design, this isn’t the only way to

achieve this functionality. We could use ARIA attributes to
associate the hint with the input:
<div class="field">
<label for="password">Password</label>
<p class="field-hint" id="passwordhint">Must contain 8+
characters with at least 1 number and 1 uppercase letter.</p>
<input type="password" id="password" name="password"
aria-describedby="passwordhint">
</div>
The aria-describedby attribute is used to connect the

hint by its id — just like the for attribute for labels, but
in reverse. It’s appended to the control’s label and read out
after a short pause. In this example, “password [pause] must
contain eight plus characters with at least one number and

one uppercase letter.”
There are other differences too. First, clicking the hint (a

<p> in this case) won’t focus the control, which reduces
the hit area. Second, despite ARIA’s growing support, it’s
never going to be as well supported as native elements. In
this particular case, Internet Explorer 11 doesn’t support
aria-describedby.4 This is why the first rule of ARIA is
not to use ARIA:5
If you can use a native HTML element or attribute with the seman-
“ tics and behaviour you require already built in, instead of re-pur-
posing an element and adding an ARIA role, state or property to
make it accessible, then do so.
Float Labels
The float label pattern by Matt Smith is a technique that
uses the label as a placeholder.6 The label starts inside the
control, but floats above the control as the user types, hence
the name. This technique is often lauded for its quirky, min-
imalist, and space-saving qualities.
4 http://smashed.by/arialabelinput
5 http://smashed.by/firstrule
6 http://smashed.by/floatlabel
26 Chapter 1
The float label pattern. On the left, an unfocused text field shows the label
inside; on the right, when the text field receives focus, the label moves above
the field.
Unfortunately, there are several problems with this

approach. First, there is no space for a hint because the label
and hint are one and the same. Second, they’re hard to read,
due to their poor contrast and small text, as they’re typically
designed. (Lower contrast is necessary so that users have
a chance to differentiate between a real value and a place-
holder.) Third, like placeholders, they may be mistaken for a
value and could get cropped.
And float labels don’t actually save space. The label needs
space to move into in the first place. Even if they did save
space, that’s hardly a good reason to diminish the usability
of forms.
Seems like a lot of effort when you could simply put labels above
“ inputs & get all the benefits/none of the issues.7

— Luke Wroblewski on float labels
7 http://smashed.by/luketweet
Quirky and minimalist interfaces don’t make users feel

awesome — obvious, inclusive, and robust interfaces do.
Artificially reducing the height of forms like this is both
uncompelling and problematic.
Instead, you should prioritize making room for an ever-

present, readily available label (and hint if necessary) at
the start of the design process. This way you won’t have to
squeeze content into a small space.
We’ll be discussing several, less artificial techniques to

reduce the size of forms shortly.
The Question Protocol

One powerful and natural way to reduce the size of a form
is to use a question protocol.8 It helps ensure you know why
you are asking every question or including a form field.
Does the registration form need to collect first name, last

name, email address and password? Are there better or
alternative ways to ask for this information that simplify
the experience?
8 http://smashed.by/questionprotocol
28 Chapter 1
In all likelihood, you don’t need to ask for the user’s first and
last name for them to register. If you need that information
later, for whatever reason, ask for it then. By removing these
fields, we can halve the size of the form. All without resort-
ing to novel and problematic patterns.
NO PASSWORD SIGN-IN
One way to avoid asking users for a password is to use the
no password sign-in pattern. It works by making use of the
security of email (which already needs a password). Users
enter only their email address, and the service sends a
special link to their inbox. Following it logs the user into the
service immediately.

Medium’s passwordless sign-in screen.
Not only does this reduce the size of the form to just one
field, but it also saves users having to remember another
password. While this simplifies the form in isolation, in
other ways it adds some extra complexity for the user.
First, users might be less familiar with this approach, and

many people are worried about online security. Second,
having to move away from the app to your email account is
long-winded, especially for users who know their password,
or use a password manager.
It’s not that one technique is always better than the other.
It’s that a question protocol urges us to think about this as
part of the design process. Otherwise, you’d mindlessly add
a password field on the form and be done with it.
PASSPHRASES
Passwords are generally short, hard to remember, and easy
to crack. Users often have to create a password of more than
eight characters, made up of at least one uppercase and one
lowercase letter, and a number. This micro-interaction is
hardly ideal.
Sorry but your password must contain an uppercase letter, a num-
“ ber, a haiku, a gang sign, a hieroglyph, and the blood of a virgin.

— Anonymous internet meme
30 Chapter 1
Instead of a password, we could ask users for a passphrase.9

A passphrase is a series of words such as “monkeysin-
mygarden” (sorry, that’s the first thing that comes to mind).
They are generally easier to remember than passwords, and
they are more secure owing to their length — passphrases
must be at least 16 characters long.
The downside is that passphrases are less commonly used

and, therefore, unfamiliar. This may cause anxiety for users
who are already worried about online security.
Whether it’s the no password sign-in pattern or pass-

phrases, we should only move away from convention once
we’ve conducted thorough and diverse user research. You
don’t want to exchange one set of problems for another
unknowingly.
Field Styling
The way you style your form components will, at least in part,
be determined by your product or company’s brand. Still, label
position and focus styles are important considerations.
9 http://smashed.by/userfriendlypw
LABEL POSITION
Matteo Penzo’s eye-tracking tests showed that position-
ing the label above (as opposed to beside) the form control
works best.10
Placing a label right over its input field permitted users to capture
“ both elements with a single eye movement.
But there are other reasons to put the label above the field.
On small viewports there’s no room beside the control. And
on large viewports, zooming in increases the chance of the
text disappearing off screen.11
Also, some labels contain a lot of text, which causes it to

wrap onto multiple lines, which would disrupt the visual
rhythm if placed next to the control. While you should aim
to keep labels terse, it’s not always possible. Using a pattern
that accommodates varying content — by positioning labels
above the control — is a good strategy.
LOOK, SIZE, AND SPACE

Form fields should look like form fields. But what does that
mean exactly?
10 http://smashed.by/labelplacement
11 http://smashed.by/nofloatreasons
32 Chapter 1
It means that a text box should look like a text box. Empty
boxes signify “fill me in” by virtue of being empty, like a
coloring-in book. This happens to be part of the reason
placeholders are unhelpful. They remove the perceived
affordance an empty text box would otherwise provide.
This also means that the empty space should be boxed in

(bordered). Removing the border, or having only a bottom
border, for example, removes the perceived affordances. A
bottom border might at first appear to be a separator. Even
if you know you have to fill something in, does the value go
above the line or below it?
Spatially, the label should be closest to its form control,

not the previous field’s control. Things that appear close
together suggest they belong together.12 Having equal
spacing might improve aesthetics, but it would be at the
cost of usability.
Finally, the label and the text box itself should be large
enough to read and tap. This probably means a font size
of at least 16 pixels, and ideally an overall tap target of at
least 44px.13
12 http://smashed.by/lawofproximity
13 http://smashed.by/touchtargetsizes
FOCUS STYLES
Focus styles are a simpler prospect. By default, browsers put
an outline around the element in focus so users, especially
those who use a keyboard, know where they are. The prob-
lem with the default styling is that it is often faint and hard
to see, and somewhat ugly.
While this is the case, don’t be tempted to remove it,

because doing so will diminish the user experience greatly
for those traversing the screen by keyboard. We can over-
ride the default styling to make it clearer and more aestheti-
cally pleasing.
input:focus {
outline: 4px solid #ffbf47;
}
The Email Field

Despite its simple appearance there are some important
details that have gone into the field’s construction which
affect the experience.
The email field.

34 Chapter 1
As noted earlier, some fields have a hint in addition to the

label, which is why the label is inside a child span. The
field-label class lets us style it through CSS.
<div class="field">
<label for="email">
<span class="field-label">Email address</span>
</label>
<input type="email" id="email" name="email">
</div>
The label itself is “Email address” and uses sentence case.

In Making A Case For Letter Case, John Saito explains that
sentence case (as opposed to title case) is generally easier to
read, friendlier, and makes it easier to spot nouns.14 Whether
you heed this advice is up to you, but whatever style you
choose, be sure to use it consistently.
The input’s type attribute is set to email, which triggers an

email-specific onscreen keyboard on mobile devices. This
gives users easy access to the @ and . (dot) symbols which
every email address must contain.
14 http://smashed.by/lettercase
Android’s onscreen keyboard for the email field.
People using a non-supporting browser will see a standard

text input (<input type="text">). This is a form of pro-
gressive enhancement, which is a cornerstone of designing
inclusive experiences.
PROGRESSIVE ENHANCEMENT
Progressive enhancement is about users. It just happens
to make our lives as designers and developers easier too.
Instead of keeping up with a set of browsers and devices
(which is impossible!) we can just focus on features.
First and foremost, progressive enhancement is about

always giving users a reasonable experience, no matter their
browser, device, or quality of connection. When things go
wrong — and they will — users won’t suffer in that they can
still get things done.
36 Chapter 1
There are a lot of ways an experience can go wrong. Perhaps

the style sheet or script fails to load. Maybe everything
loads, but the user’s browser doesn’t recognize some HTML,
CSS, or JavaScript. Whatever happens, using progressive
enhancement when designing experiences stops users hav-
ing an especially bad time.
It starts with HTML for structure and content. If CSS or

JavaScript don’t load, it’s fine because the content is there.
If everything loads OK, perhaps various HTML elements

aren’t recognized. For example, some browsers don’t under-
stand <input type="email">. That’s fine, though, because
users will get a text box (<input type="text">) instead.
Users can still enter an email address; they just don’t get an
email-specific keyboard on mobile.
Maybe the browser doesn’t understand some fancy CSS, and

it will just ignore it. In most cases, this isn’t a problem. Let’s
say you have a button with border-radius: 10px. Browsers
that don’t recognize this rule will show a button with angled
corners. Arguably, the button’s perceived affordance is
reduced, but users are left unharmed. In other cases it might
be helpful to use feature queries.15
15 http://smashed.by/featurequeries
Then there is JavaScript, which is more complicated. When

the browser tries to parse methods it doesn’t recognize, it will
throw a hissy fit. This can cause your other (valid and sup-
ported) scripts to fail. If your script doesn’t first check that the
methods exist (feature detection) and work (feature testing)
before using them, then users may get a broken interface. For
example, if a button’s click handler calls a method that’s not
recognized, the button won’t work. That’s bad.
That’s how you enhance. But what’s better is not needing an

enhancement at all. HTML with a little CSS can give users
an excellent experience. It’s the content that counts and you
don’t need JavaScript for that. The more you can rely on con-
tent (HTML) and style (CSS), the better. I can’t emphasize
this enough: so often, the basic experience is the best and
most performant one.16 There’s no point in enhancing some-
thing if it doesn’t add value (see inclusive design principle 7).
Of course, there are times when the basic experience isn’t as

good as it could be — that’s when it’s time to enhance. But if
we follow the approach above, when a piece of CSS or Java-
Script isn’t recognized or executed, things will still work.
16 http://smashed.by/designperf
38 Chapter 1
Progressive enhancement makes us think about what hap-

pens when things fail. It allows us to build experiences with
resilience baked in. But equally, it makes us think about
whether an enhancement is needed at all; and if it is, how
best to go about it.
The Password Field

We’re using the same markup as the email field discussed
earlier. If you’re using a template language, you’ll be able to
create a component that accommodates both types of field.
This helps to enforce inclusive design principle 3, be consistent.
The password field using the hint text pattern.
<div class="field">
<label for="password">
<span class="field-label">Choose password</span>
<span class="field-hint">Must contain 8+ characters
with at least 1 number and 1 uppercase letter.</span>
</label>
<input type="password" id="password" name="password">
</div>
The password field contains a hint. Without one, users

won’t understand the requirements, which is likely to cause
an error once they try to proceed.
The type="password" attribute masks the input’s value by

replacing what the user types with small black dots. This is
a security measure that stops people seeing what you typed
if they happen to be close by.
A PASSWORD REVEAL
Obscuring the value as the user types makes it hard to fix
typos. So when one is made, it’s often easier to delete the
whole entry and start again. This is frustrating as most
users aren’t using a computer with a person looking over
their shoulder.
Owing to the increased risk of typos, some registration

forms include an additional “Confirm password” field. This
is a precautionary measure that requires the user to type the
same password twice, doubling the effort and degrading the
user experience.
Instead, it’s better to let users reveal their password, which

speaks to principles 4 and 5, give control and offer choice
respectively. This way users can choose to reveal their pass-
word when they know nobody is looking, reducing the risk
of typos.
40 Chapter 1
The password field with a “Show password” button beside it.
Recent versions of Internet Explorer and Microsoft Edge

provide this behavior natively. As we’ll be creating our own
solution, we should suppress this feature using CSS like this:
input[type=password]::-ms-reveal {
display: none;
}
Now we’re ready to enhance the interface with our own

version.
First, we need to inject a button next to the input. The

<button> element should be your go-to element for chang-
ing anything with JavaScript — except, that is, for changing
location, which is what links are for. When clicked, it should
toggle the type attribute between password and text; and
the button’s label between “Show” and “Hide.”
function PasswordReveal(input) {
// store input as a property of the instance
// so that it can be referenced in methods
// on the prototype
this.input = input;
this.createButton();
};
PasswordReveal.prototype.createButton = function() {
// create a button
this.button = $('<button type="button">Show password</
button>');
// inject button
$(this.input).parent().append(this.button);
// listen to the button’s click event
this.button.on('click', $.proxy(this, 'onButtonClick'));
};
PasswordReveal.prototype. {
// Toggle input type and button text
if(this.input.type === 'password') {
this.input.type = 'text';
this.button.text('Hide password');
} else {
this.input.type = 'password';
this.button.text('Show password');
}
};
42 Chapter 1
JavaScript Syntax and Architectural Notes
As there are many flavors of JavaScript, and different ways

in which to architect components, we’re going to walk
through the choices used to construct the password reveal
component, and all the upcoming components in the book.
First, we’re using a constructor. A constructor is a func-

tion conventionally written in upper camel case —
PasswordReveal, not passwordReveal. It’s initialized using
the new keyword, which lets us use the same code to create
several instances of the component:
var passwordReveal1 = new PasswordReveal(document.

getElementById('input1'));
var passwordReveal2 = new PasswordReveal(document
getElementById('input2'));
Second, the component’s methods are defined on the

prototype — PasswordReveal.prototype.onButtonClick
for example. The prototype is the most performant way
to share methods across multiple instances of the same
component.
Third, jQuery is being used to create and retrieve elements,

and listen to events. While jQuery may not be necessary or
preferred, using it means that this book can focus on forms
and not on the complexities of cross-browser components.
If you’re a designer who codes a little bit, then jQuery’s ubiq-

uity and low-barrier to entry should be helpful. By the same
token, if you prefer not to use jQuery, you’ll have no trouble
refactoring the components to suit your preference.
You may have also noticed the use of the $.proxy

function. This is jQuery’s implementation of
Function.prototype.bind. If we didn’t use this function
to listen to events, then the event handler would be called
in the element’s context (this). In the example above,
this.button would be undefined. But we want this to be
the password reveal object instead, so that we can access its
properties and methods.
Alternative Interface Options
The password reveal interface we constructed above toggles

the button’s label between “Show password” and “Hide pass-
word.” Some screen reader users can get confused when the
button’s label is changed; once a user encounters a button,
they expect that button to persist. Even though the button is
persistent, changing the label makes it appear not to be.
If your research shows this to be a problem, you could try

two alternative approaches.
44 Chapter 1
First, use a checkbox with a persistent label of “Show pass-

word.” The state will be signaled by the checked attribute.
Screen reader users will hear “Show password, checkbox,
checked” (or similar). Sighted users will see the checkbox
tick mark. The problem with this approach is that check-
boxes are for inputting data, not controlling the interface.
Some users might think their password will be revealed to
the system.
Or, second, change the button’s state — not the label. To

convey the state to screen reader users, you can switch the
aria-pressed attribute between true (pressed) and false
(unpressed).
<button type="button" aria-pressed="true">

Show password
</button>
When focusing the button, screen readers will announce,

“Show password, toggle button, pressed” (or similar). For
sighted users, you can style the button to look pressed or
unpressed accordingly using the attribute selector like this:
[aria-pressed="true"] {
box-shadow: inset 0 0 0 0.15rem #000, inset 0.25em 0.25em
0 #fff;
}
Just be sure that the unpressed and pressed styles are

obvious and differentiated, otherwise sighted users may
struggle to tell the difference between them.
MICROCOPY
The label is set to “Choose password” rather than “Pass-
word.” The latter is somewhat confusing and could prompt
the user to type a password they already possess, which
could be a security issue. More subtly, it might suggest the
user is already registered, causing users with cognitive
impairments to think they are logging in instead.
Where “Password” is ambiguous, “Choose password” pro-

vides clarity.
Button Styles
What’s a button? We refer to many different types of
components on a web page as a button. In fact, I’ve already
covered two different types of button without calling them
out. Let’s do that now.
Buttons that submit forms are “submit buttons” and they

are coded typically as either <input type="submit"> or
<button type="submit">. The <button> element is more
malleable in that you can nest other elements inside it.
46 Chapter 1
But there’s rarely a need for that. Most submit buttons con-
tain just text.
Note: In older versions of Internet Explorer, if you have

multiple <button type="submit">s, the form will submit
the value of all the buttons to the server, regardless of which
was clicked.17 You’ll need to know which button was clicked
so you can determine the right course of action to take,
which is why this element should be avoided.
Other buttons are injected into the interface to enhance the

experience with JavaScript — much like we did with the
password reveal component discussed earlier. That was also
a <button> but its type was set to button (not submit).
In both cases, the first thing to know about buttons is that

they aren’t links. Links are typically underlined (by user agent
styles) or specially positioned (in a navigation bar) so they are
distinguishable among regular text.
When hovering over a link, the cursor will change to a

pointer. This is because, unlike buttons, links have weak
perceived affordance.18
17 http://smashed.by/submitbuttons
18 http://smashed.by/perceivedaffordance
In Resilient Web Design, Jeremy Keith discusses the idea of

material honesty.19 He says: “One material should not be
used as a substitute for another. Otherwise the end result
is deceptive.” Making a link look like a button is materially
dishonest. It tells users that links and buttons are the same
when they’re not.
Links can do things buttons can’t do. Links can be opened in a

new tab or bookmarked for later, for example. Therefore, but-
tons shouldn’t look like links, nor should they have a pointer
cursor. Instead, we should make buttons look like buttons,
which have naturally strong perceived affordance. Whether
they have rounded corners, drop shadows, and borders is up
to you, but they should look like buttons regardless.
Buttons can still give feedback on hover (and on focus) by

changing the background colour, for example.
PLACEMENT
Submit buttons are typically placed at the bottom of the
form: with most forms, users fill out the fields from top to
bottom, and then submit. But should the button be aligned
left, right or center? To answer this question, we need to
think about where users will naturally look for it.
19 https://resilientwebdesign.com/
48 Chapter 1
Field labels and form controls are aligned left (in left-to-
right reading languages) and run from top to bottom.
Users are going to look for the next field below the last one.
Naturally, then, the submit button should also be positioned
in that location: to the left and directly below the last field.
This also helps users who zoom in, as a right-aligned button
could more easily disappear off-screen.
TEXT
The button’s text is just as important as its styling. The
text should explicitly describe the action being taken. And
because it’s an action, it should be a verb. We should aim to
use as few words as possible because it’s quicker to read. But
we shouldn’t remove words at the cost of clarity.
The exact words can match your brand’s tone of voice, but
don’t exchange clarity for quirkiness.
Simple and plain language is easy for everyone to under-

stand. The exact words will depend on the type of service.
For our registration form “Register” is fine, but depending on
your service “Join” or “Sign up” might be more appropriate.
Validation
Despite our efforts to create an inclusive, simple, and friction-
free registration experience, we can’t eliminate human
error. People make mistakes and when they do, we should
make fixing them as easy as possible.
When it comes to form validation, there are a number of

important details to consider. From choosing when to give
feedback, through to how to display that feedback, down
to the formulation of a good error message — all of these
things need to be taken into account.
HTML5 VALIDATION
HTML5 validation has been around for a while now. By
adding just a few HTML attributes, supporting browsers
will mark erroneous fields when the form is submitted.
Non-supporting browsers fall back to server-side validation.
Normally I would recommend using functionality that the

browser provides for free because it’s often more perfor-
mant, robust, and accessible. Not to mention, it becomes
more familiar to users as more sites start to use the stan-
dard functionality.
50 Chapter 1
While HTML5 validation support is quite good, it’s not

implemented uniformly.20 For example, the required attri-
bute can mark fields as invalid from the outset, which isn’t
desirable. Some browsers, such as Firefox 45.7, will show
an error of “Please enter an email address” even if the user
entered something in the box, whereas Chrome, for exam-
ple, says “Please include an ‘@’ in the email address,” which
is more helpful.
We also want to give users the same interface whether

errors are caught on the server or the client. For these rea-
sons we’ll design our own solution. The first thing to do is
turn off HTML5 validation:
<form novalidate>
HANDLING SUBMISSION
When the user submits the form, we need to check if there
are errors. If there are, we need to prevent the form from
submitting the details to the server.
20 http://smashed.by/formvalidation
function FormValidator(form) {
form.on('submit', $.proxy(this, 'onSubmit'));
}
FormValidator.prototype. {
if(!this.validate()) {
e.preventDefault();
// show errors
}
};
Note that we are listening to the form’s submit event, not the
button’s click event. The latter will stop users being able to
submit the form by pressing Enter when focus is within one
of the fields. This is also known as implicit form submission.21
DISPLAYING FEEDBACK
It’s all very well detecting the presence of errors, but at this
point users are none the wiser. There are three disparate
parts of the interface that need to be updated. We’ll talk
about each of those now.
Document Title
The document’s <title> is the first part of a web page to be

read out by screen readers. As such, we can use it to quickly
inform users that something has gone wrong with their
submission.
21 http://smashed.by/implicitsubmission
52 Chapter 1
This is especially useful when the page reloads after a server

request.
Even though we’re enhancing the user experience by

catching errors on the client with JavaScript, not all errors
can be caught this way. For example, checking that an email
address hasn’t already been taken can only be checked on
the server. And in any case, JavaScript is prone to failure so
we can’t solely rely on its availability.22
Where the original page title might read “Register for

[service],” on error it should read “(2 errors) Register for
[service]” (or similar). The exact wording is somewhat down
to opinion.
The following JavaScript updates the title:
document.title = "(" + this.errors.length + ")"

+ document.title;
As noted above, this is primarily for screen reader users, but

as is often the case with inclusive design, what helps one set
of users helps everyone else too. This time, the updated title
acts as a notification in the tab.
22 http://smashed.by/everyonehasjs
The browser tab title prefixed with “(2 errors)” acting as a quasi notification.
Error Summary
In comparison with the title element, the error summary is

more prominent, which tells sighted users that something
has gone wrong. But it’s also responsible for letting users
understand what’s gone wrong and how to fix it.
It’s positioned at the top of the page so users don’t have to

scroll down to see it after a page refresh (should an error get
caught on the server). Conventionally, errors are colored red.
However, relying on color alone could exclude colorblind
users. To draw attention to the summary, consider also
using position, size, text, and iconography.
The panel includes a heading, “There’s a problem,” to indi-

cate the issue. Notice it doesn’t say the word “Error,” which
isn’t very friendly. Imagine you were filling out your details
to purchase a car in a showroom and made a mistake. The
salesperson wouldn’t say “Error” — in fact it would be odd if
they did say that.
54 Chapter 1
Error summary panel positioned toward the top of the screen.
<div class="errorSummary" role="group" tabindex="-1"

aria-labelledby="errorSummary-heading">
<h2 id="errorSummary-heading">There’s a problem</h2>
<ul>
<li><a href="#emailaddress">Enter an email address
</a></li>
<li><a href="#password">The password must contain an
uppercase letter</a></li>
</ul>
</div>
The container has a role of group, which is used to group

a set of interface elements: in this case, the heading and the
error links. The tabindex attribute is set to -1, so it can be
focused programmatically with JavaScript (when the form is
submitted with mistakes). This ensures the error summary
panel is scrolled into view. Otherwise, the interface would
appear unresponsive and broken when submitted.
Note: Using tabindex="0" means it will be permanently

focusable by way of the Tab key, which is a 2.4.3 Focus Order
WCAG fail. If users can tab to something, they expect it will
actually do something.
FormValidator.prototype.showSummary = function () {
// ...
this.summary.focus();
};
Underneath, there’s a list of error links. Clicking a link will

set focus to the erroneous field, which lets users jump into
the form quickly. The link’s href attribute is set to the con-
trol’s id, which in some browsers is enough to set focus to it.
However, in other browsers, clicking the link will just scroll
the input into view, without focusing it. To fix this we can
focus the input explicitly.
e.preventDefault();
var href = e.target.href;
var id = href.substring(href.indexOf("#"), href.length);
$(id).focus();
};
56 Chapter 1
When there aren’t any errors, the summary panel should be

hidden. This ensures that there is only ever one summary
panel on the page, and that it appears consistently in the
same location whether errors are rendered by the client or the
server. To hide the panel we need to add a class of hidden.
<div class="errorSummary hidden" ...></div>
.hidden {
display: none;
}
You could use the hidden attribute/property to toggle an

element’s visibility, but there’s less support for it. Inclusive
design is about making decisions that you know are unlikely
to exclude people. Using a class aligns with this philosophy.
Inline Errors
We need to put the relevant error message just above the

field. This saves users scrolling up and down the page to
check the error message, and keeps them moving down
the form. If the message was placed below the field, we’d
increase the chance of it being obscured by the browser
autocomplete panel or by the onscreen keyboard.23
23 http://smashed.by/errormessages
Inline error pattern with red error text and warning icon just above the field.
<div class="field">
<label for="blah">
<span class="field-error">
<svg width="1.5em" height="1.5em">
<use xmlns:xlink="http://www.w3.org/1999/xlink"
xlink:href="#warning-icon"></use></svg>
Enter your email address.
</span>
<span class="field-error">Enter an email address</span>
</label>
</div>
Like the hint pattern mentioned earlier, the error message

is injected inside the label. When the field is focused,
screen reader users will hear the message in context, so
they can freely move through the form without having to
refer to the summary.
The error message is red and uses an SVG warning icon to

draw users’ attention. If we’d used only a color change to
denote an error, this would exclude color-blind users. So this
works really well for sighted users — but what about screen
reader users?
58 Chapter 1
To give both sighted and non-sighted users an equivalent

experience, we can use the well-supported aria-invalid
attribute. When the user focuses the input, it will now
announce “Invalid” (or similar) in screen readers.
<input aria-invalid="false">
Note: The registration form only consists of text inputs. In

chapter 3, “A Flight Booking Form,” we’ll look at how to inject
errors accessibly for groups of fields such as radio buttons.
SUBMITTING THE FORM AGAIN

When submitting the form for a second time, we need to
clear the existing errors from view. Otherwise, users may
see duplicate errors.
this.resetPageTitle();
this.resetSummaryPanel();
this.removeInlineErrors();
if(!this.validate()) {
e.preventDefault();
this.updatePageTitle();
this.showSummaryPanel();
this.showInlineErrors();
}
};
INITIALIZATION
Having finished defining the FormValidator component,
we’re now ready to initialize it. To create an instance of
FormValidator, you need to pass the form element as the
first parameter.
var validator = new FormValidator(document.

getElementById('registration'));
To validate the email field, for example, call the
addValidator() method:
validator.addValidator('email', [{
method: function(field) {
return field.value.trim().length > 0;
},
message: 'Enter your email address.'
},{
method: function(field) {
return (field.value.indexOf('@') > -1);
},
message: 'Enter the ‘at’ symbol in the email address.'
}]);
The first parameter is the control’s name, and the second is

an array of rule objects. Each rule contains two properties:
method and message. The method is a function that tests
various conditions to return either true or false. False puts
the field into an error state, which is used to populate the
interface with errors as discussed earlier.
60 Chapter 1
Forgiving Trivial Mistakes
In The Design of Everyday Things, Don Norman talks about

designing for error. He talks about the way people converse:
If a person says something that we believe to be false, we question
“ and debate. We don’t issue a warning signal. We don’t beep. We

don’t give error messages. […] In normal conversations between two
friends, misstatements are taken as normal, as approximations to
what was really meant.
Unlike humans, machines are not intelligent enough to

determine the meaning of most actions, but they are often
far less forgiving of mistakes than they need to be. Jared
Spool makes a joke about this in “Is Design Metrically
Opposed?” (about 42 minutes in):24
It takes one line of code to take a phone number and strip out all the
“ dashes and parentheses and spaces, and it takes ten lines of code to
write an error message that you left them in.
The addValidator method (shown above) demonstrates

how to design validation rules so they forgive trivial mis-
takes. The first rule, for example, trims the value before
checking its length, reducing the burden on the user.
24 http://smashed.by/onelineofcode
LIVE INLINE VALIDATION

Live inline validation gives users feedback as they type or
when they leave the field (onblur). There’s some evidence
to show that live inline validation improves accuracy and
decreases completion times in long forms.25 This is partially
to do with giving users feedback when the field’s require-
ments are fresh in users’ minds. But live inline validation (or
live validation for short) poses several problems.
For entries that require a certain number of characters, the

first keystroke will always constitute an invalid entry. This
means users will be interrupted early, which can cause them
to switch mental contexts, from entering information to
fixing it.
Alternatively, we could wait until the user enters enough

characters before showing an error. But this means users
only get feedback after they have entered a correct value,
which is somewhat pointless.
We could wait until the user leaves the field (onblur), but
this is too late as the user has mentally prepared for (and
often started to type in) the next field. Moreover, some users
switch windows or use a password manager when using a
form. Doing so will trigger the blur event, causing an error
to show before the user is finished. All very frustrating.
25 http://smashed.by/inlinevalidation
62 Chapter 1
Remember, there’s no problem with giving users feedback

without a page refresh. Nor is there a problem with putting
the error messages inline (next to fields) — we’ve done this
already. The problem with live feedback is that it interrupts
users either too early or too late, which often results in a
jarring experience.
If users are seeing errors often, there’s probably something

wrong elsewhere. Focus on shortening your form and
providing better guidance (good labeling and hint text). This
way users shouldn’t see more than the odd error. We’ll look
at longer forms in the next chapter.
CHECKLIST AFFIRMATION PATTERN

A variation of live validation involves ticking off rules
(marking them as complete) as the user types. This is less
invasive than live validation but isn’t suited to every type
of field. Here’s an example of MailChimp’s sign-up form,
which employs this technique for the password field.
MailChimp’s password field with instructions that get marked as the user
meets the requirements.
You should put the rules above the field. Otherwise the
onscreen keyboard could obscure the feedback. As a result,
users may stop typing and hide the keyboard to then check
the feedback.
A NOTE ON DISABLING SUBMIT BUTTONS

Some forms are designed to disable the submit button
until all the fields become valid. There are several prob-
lems with this.
First, users are left wondering what’s actually wrong with

their entries. Second, disabled buttons are not focusable,
which makes it hard for the button to be discovered by blind
users navigating using the Tab key. Third, disabled buttons
are hard to read as they are grayed out.
As we’re providing users with clear feedback, when the user

expects it, there’s no good reason to take control away from
the user by disabling the button anyway.
CRAFTING A GOOD ERROR MESSAGE

There’s nothing more important than content. Users don’t
come to your website to enjoy the design. They come to
enjoy the content or the outcome of using a service.
64 Chapter 1
Even the most thought-out, inclusive and beautifully

designed experience counts for nothing if we ignore the
words used to craft error messages. One study showed that
showing custom error messages increased conversions
by 0.5% which equated to more than £250,000 in yearly
revenue.26
“ Content is the user experience.

— Ginny Redish
Like labels, hints, and any other content, a good error mes-
sage provides clarity in as few words as possible. Normally,
we should drive the design of an interface based on the
content — not the other way around. But in this case, under-
standing how and why you show error messages influences
the design of the words. This is why Jared Spool says “con-
tent and design are inseparable work partners.”27
We’re showing messages in the summary at the top of the

screen and next to the fields. Maintaining two versions of
the same message is a hard sell for an unconvincing gain.
Instead, design an error message that works in both places.
“Enter an ‘at’ symbol” needs context from the field label
to make sense. “Your email address needs an ‘at’ symbol”
works well in both places.
26 http://smashed.by/errormessagesroi
27 http://smashed.by/contentdesign
Avoid pleasantries, like starting each error message with

“Please.” On the one hand, this seems polite; on the other, it
gets in the way and implies a choice.
Whatever approach you take, there’s going to be some rep-

etition due to the nature of the content. And testing usually
involves submitting the form without entering any infor-
mation at all. This makes the repetition glaringly obvious,
which may cause us to flip out. But how often is this the
case? Most users aren’t trying to break the interface.
An error summary containing a wall of error messages makes the

beginning of the words seem too repetitive.
Different errors require different formatting. Instructions

like “Enter your first name” are natural. But “Enter a first
name that is 35 characters or less” is longer, wordier, and
less natural than a description like “First name must be 35
characters or less.”
66 Chapter 1
Here’s a checklist:
• Be concise. Don’t use more words than are necessary,

but don’t omit words at the cost of clarity.
• Be consistent. Use the same tone, the same words, and
the same punctuation throughout.
• Be specific. If you know why something has gone
wrong, say so. “The email is invalid.” is ambiguous and
puts the burden on the user. “The email needs an ‘at’
symbol” is clear.
• Be human, avoid jargon. Don’t use words like invalid,
forbidden, and mandatory.
• Use plain language. Error messages are not an oppor-
tunity to promote your brand’s humorous tone of voice.
• Use the active voice. When an error is an instruction
and you tell the user what to do. For example, “Enter
your name,” not “First name must be entered.”
• Don’t blame the user. Let them know what’s gone
wrong and how to fix it.
Summary
In this chapter we solved several fundamental form design
challenges that are applicable well beyond a simple registra-
tion form. In many respects, this chapter has been as much
about what not to do, as it has about what we should. By
avoiding novel and artificial space-saving patterns to focus
on reducing the number of fields we include, we avoid sev-
eral usability failures while simultaneously making forms
more pleasant.
THINGS TO AVOID
• Using the placeholder attribute as a mechanism for

storing label and hint text.
• Using incorrect input types.
• Styling buttons and links the same.
• Validating fields as users type.
• Disabling submit buttons.
• Using complex jargon and brand-influenced microcopy.
DEMOS
• Registration form: http://smashed.by/regformdemo

68 Chapter 2
A Checkout Form
Here’s what I bring to the table: a valid credit card, 90 seconds of my
“ time, and my right thumb. The rest is up to you.

— Melanie Jones, “8 things parenting taught me about accessibility”1
In 2008 I worked on Boots.com, where we designed a

single-page checkout flow. This involved the trendiest
of techniques from that era, including accordions, Ajax
and client-side validation. Each step of the flow (delivery
address, delivery options, payment) was an accordion panel,
submitted via Ajax. On successful submission, the panel
collapsed and the next one opened.
The Boots accordion single page checkout flow.
1 http://smashed.by/parenting
A Checkout Form 69
Users struggled to complete their orders. Errors were hard to

fix as users had to scroll up and down. And the accordion was
a distraction. Inevitably, Boots asked us to make changes. We
redesigned it so each panel became a page removing the need
for an accordion and Ajax. (We kept the client-side validation
to avoid an unnecessary trip to the server.)
The Boots one-thing-per-page checkout flow.
This converted a lot better. Although I can’t remember

the exact numbers (it was back in 2008, remember!), the
client was happy with the results. Six years later, in 2014,
at Just Eat, the same thing happened. We redesigned the
single-page checkout flow so each section became a page.
This time I noted the numbers. The result was a whopping
5% increase in conversion. This equated to 2 million orders a
year. That’s orders, not revenue.
70 Chapter 2
The Just Eat one-thing-per-page checkout mobile screens.
Two years later, in 2016, Robin Whittleton from the UK’s Gov-
ernment Digital Service (GDS), told me that putting each thing
on a page of its own was a design pattern called “one thing
per page.”2 Behind the improved numbers, there are many
reasons why it drastically improves the user experience.
One Thing Per Page

One thing per page is about splitting up a complicated process
into small chunks and placing them on screens of their own.
For example, instead of putting delivery address, delivery
options, and payment forms on one page, we put them on
separate pages.
2 http://smashed.by/1thingperpage
A Checkout Form 71
It’s not necessarily about having one element or component

on a page (although it could be). In all likelihood, you’ll still
have a header. Similarly, it’s not about having a single form
field on each page (although it absolutely could). And it
doesn’t mean you’ll always end up with one question per
page either.
Forms expert Caroline Jarrett, who first wrote about the

pattern in 2015, explains that user research “will quickly
show you that some questions will be best grouped into a
longer page.”3
However, she also explains that “questions that naturally ‘go

together’ from the point of view of designers […] don’t need
to be on the same page to work for users.” She provides an
enlightening example when, for GOV.UK Verify, they tested
putting “Create a username” on one page, and “Create a
password” on the next.
Like most designers, Caroline thought that putting them on

separate pages would be overkill. In reality, users weren’t
bothered. So start with one thing (field or question) per
page, then, through research, find out if grouping fields
improves the experience.
3 http://smashed.by/nomoreaccordions
72 Chapter 2
While this pattern often bears wonderful and delicious fruit

(or orders and conversions, if you hate my analogies), it’s
useful to understand why it works so well.
• Inclusive design principle 6 says we should “design interfac-

es that help users focus on core tasks by prioritizing them.” It
even goes on to say that “people should be able to focus on
one thing at a time.” One thing per page simply follows
this principle to the letter, and in doing so drastically
reduces the cognitive burden on users.
• When users fill in a small form, “errors are caught and

shown early” and often. If there’s one thing to fix, it’s
easy to fix, which reduces the chance of users giving up
on the task.
• If pages have little on them, they’ll load quickly. Faster

pages reduce the risk of users leaving and they build trust.
• By submitting information frequently, we can save user

information in a more granular fashion. If a user drops
out, we can, for example, send them an email prompt-
ing them to complete their order.
• Conversely, a long form increases the chance of a page

timing out, or the computer freezing. This is what hap-
pens to Daniel, the lead character in Ken Loach’s film,
I, Daniel Blake.4 With declining health and having never
4 http://smashed.by/danielblake
A Checkout Form 73
used a computer, it freezes and he loses his data. In the

end, he gives up.
• It adds a sense of progression and increases momen-
tum because the user is constantly moving forwards
step by step.
• It lets you design interfaces that capitalize on maximal
screen space — interfaces that wouldn’t work so well if
part of a larger form. We’ll see an example of this in the
next chapter, when we design an airplane seat chooser.
Flow and Order

In Forms That Work, Caroline Jarrett and Gerry Gaffney explain
the importance of asking questions in a sensible order:5
Asking for information at the wrong time can alienate a user. The
“ same question put at the right moment can be entirely acceptable.

Think about buying a car. You’re just browsing, getting a sense of
what is available. A salesperson comes along and starts to ask you
how you’ll pay. Would you answer? Or would you think, “If that
person doesn’t stop annoying me, I’m out of here”?
Now think about the point where you’ve told the salesper-
son which car you want to buy. Now it’s appropriate to start
negotiating about payment. It would be quite odd if the
salesperson did not do so.
5 http://smashed.by/formsthatwork
74 Chapter 2
Just like the car salesperson, we’ll ask for the right information
at the right time. For example, payment happens toward the
end. Users will be given a chance to check their order before
submitting it. Finally, the confirmation page acts as a sales
receipt for administrative purposes. Here’s the complete flow:
1. Email address
2. Mobile phone (optional)
3. Delivery address
4. Delivery options
5. Delivery notes
6. Payment
7. Check your answers
8. Confirmation
Guest Checkout
Inclusive design principle 2, “Consider situation,” says:
People use your interface in different situations. Make sure your
“ interface delivers a valuable experience to people regardless of their

circumstances.
We’ll first design the checkout journey for anonymous

users. Not letting users check out as a guest is just about the
worst thing we can do, as Jared Spool attests to beautifully
in “The $300 Million Button.”6
6 http://smashed.by/3mbutton
A Checkout Form 75
The article tells a story of one company losing 300 million

dollars because they thought forcing users to register first
would help speed up subsequent purchases. While this is
true, it also assumes users want to sign up in the first place.
Bearing in mind what we’ve learned about flow and order,

and the question protocol, that story is hardly a surprise.
What value does a first-time user get in return for the added
effort of signing up? Nothing. And that’s all you really need
to know.
We’ll look at a more appropriate time to ask users to sign

up, as well as looking at how to optimize the journey for
second-time users later on in the chapter.
1. Email Address
In chapter 1, “A Registration Form,” we had to ask users for
an email address (see page 33). We can reuse that pattern
here, saving us the effort of solving the same problem again
from scratch.
There is, however, an opportunity to adapt the content to fit

this context better. By that, I mean users may wonder why
they’re being asked for an email address just to purchase
something. One of the main takeaways from chapter 1 was
the need to justify the existence of each and every form field.
76 Chapter 2
The email address field with hint text explaining why users
are being asked for this.
Here, it’s because we can send users a receipt, which is par-

ticularly important if checking out anonymously. Addition-
ally, the email may provide details about how to return the
item, or cancel or change the order. We can tell users this
transparently via the hint text.
Note: The button’s positioning and styling is the same

as set out in “A Registration Form.” But the label is set to
“Continue,” which implies progress, and is better suited to
the linear checkout flow.
2. Mobile Phone
Like the email field, we should be asking ourselves why
we’re requesting a phone number. We know the courier
offers real-time text messages on the day of delivery —
A Checkout Form 77
but the customer doesn’t. So we tell them via the hint.

Remember, the hint is not just for formatting rules: it’s for
anything that will help users fill out the field. This transpar-
ency builds trust, reduces effort, and promotes the feature
all at the same time.
Mobile phone field with “(optional)” text at the end of the label.
<div class="field">
<label for="mobile">
<span class="field-label">Mobile number (optional)</span>
<span class="field-hint">So we can notify you about
delivery</span>
</label>
<input type="tel" id="mobile" name="mobile">
</div>
The input’s type="tel" attribute will spawn a tele-

phone-specific keyboard on mobile devices. This makes it
easier to enter a phone number thanks to the larger keypad.
78 Chapter 2
The iOS on-screen keyboard for the telephone input.
MARKING OPTIONAL OR REQUIRED FIELDS

While real-time notifications add value, we shouldn’t assume
everyone wants to receive them, nor that everyone has a mobile
phone. So we let users choose to skip this field by marking the
field as optional. This way, users can opt in if they like.
By convention, required fields are marked with an aster-

isk. A legend is usually placed above the form to denote its
meaning, but as Luke Wroblewski says:7
[…] including the phrase “optional” after a label is much clearer than
“ any visual symbol you could use to mean the same thing. Someone
may always wonder ‘what does this asterisk mean?’ and have to go
hunting for a legend that explains things.
7 http://smashed.by/requiredfields
A Checkout Form 79
You might also be wondering why we’re marking optional

fields, instead of required ones. In “Required Versus
Optional Fields,” Jessica Enders says:8
[…] think about what we are doing when we mark something in an
“ interface. We are trying […] to indicate that it’s different.
Thanks to the question protocol, most fields should be

required, so we mark optional fields instead.
Note: Putting the aria-required="true" attribute on an

input will tell screen reader users that the field is required.
As we’re only marking optional fields, this attribute isn’t
necessary. And the optional text from within the label will
be announced by screen readers too, giving sighted and
non-sighted users an equivalent experience.
3. Delivery Address
The delivery address contains five fields that together make
up an address. Visually there is a slight difference between
the fields: field width.
8 http://smashed.by/optfields
80 Chapter 2
Delivery address fields: address line 1, line 2, city, and postcode.
<div class="field">
<label for="address1">
<span class="field-label">Address line 1</span>
</label>
<input type="text" id="address1" name="address1">
</div>
<div class="field">
<label for="address2">
<span class="field-label">Address line 2 (optional)</span>
</label>
<input type="text" id="address2" name="address2">
</div>
A Checkout Form 81
<div class="field">
<label for="city">
<span class="field-label">City</span>
</label>
<input type="text" id="city" name="city">
</div>
<div class="field">
<label for="postcode">
<span class="field-label">Postcode</span>
</label>
<input type="text" id="postcode" name="postcode">
</div>
FIELD WIDTH
In “Write Less Damn Code”, Heydon Pickering jokingly
points out that the reason some people used to add XHTML
1.1 compliant banners to their website was to ensure the
height of the menu matches the height of the content.9
Similarly, you might be tempted to give every address field
the same width.
But giving the postcode field the same width as every other
field increases the cognitive effort needed to fill it out. This
is because the width gives users a clue as to the length of
the content it requires.
9 http://smashed.by/heydontalk
82 Chapter 2
Baymard Institute’s study10 found that:
If a field was too long or too short, [users] started to wonder if they had
“ misunderstood the label. […] This was especially true for fields with
uncommon data or a technical label like CVV [card verification value].
As postcodes consist of six to eight characters, the field’s

width should be smaller than the other fields. You should
apply this rule to every field where the length of the content
is known.
CAPTURE+ ENHANCEMENT
Capture+ is a third-party plugin that lets users search for
their address quickly and accurately.11 Instead of manually
typing each part of the address in five separate boxes, users
type into just one.
As the user types the first line of their address, suggestions

will appear from which they can select. This reduces the
number of keystrokes and therefore the chance of typos.
10 http://smashed.by/formfieldux
11 http://smashed.by/addresscapture
A Checkout Form 83
A text box using the Capture+ plugin showing options as users type
their postcode.
If no address is found, users can change the interface back

to the original address form. In doing so, we conform to
inclusive design principle 5, “Offer choice.”
Capture+ has a third-party script which you can include on

your page. But most third-party scripts don’t account for the
broad range of interaction preferences, usability, and acces-
sibility considerations. We’ll look at all of this in the next
chapter, when we build our own accessible autocomplete
component from scratch.
84 Chapter 2
4. Delivery Options
This is the first field that consists of multiple controls; in
this case, radio buttons.
Delivery option radio buttons with two options: free delivery, and premium.
<fieldset class="field">
<legend>
<span class="field-legend">Delivery options</span>
</legend>
<div class="field-radioButton">
<input type="radio" name="option" id="option"
value="Standard" checked>
<label for="option">Standard (Free, 2-3 days)</label>
</div>
<input type="radio" name="option" id="option2"
value="Premium">
<label for="option2">Premium (£6, Next day)</label>
</div>
</fieldset>
A Checkout Form 85
GROUPING
To group multiple controls, we must wrap them in a
fieldset. The legend describes the group like a label
describes the individual control.
Some screen readers, such as NVDA, will read the legend

out, along with the first individual radio button’s label
when entering the field (in either direction). In this exam-
ple, “Delivery options, Standard (Free two to three days)” is
announced. In other screen readers, such as Voiceover with
Safari, the legend is announced for every field.
If we omitted the fieldset and legend elements, screen

reader users would only hear “Standard (Free, two to three
days),” which is less clear.
You may be tempted to group all fields this way. For exam-
ple, the address form from earlier could be wrapped inside a
fieldset with a legend set to “Address.” While this is tech-
nically valid, it’s unnecessary and verbose, as the field labels
make sense without a legend. Put another way, users don’t
need to hear “Address: Address Line 1” as it doesn’t add value.
86 Chapter 2
SMART DEFAULTS
As most users will want free delivery, that option comes
first. It’s also selected by default thanks to the checked
attribute. This stops users from ever seeing an error and
gives users less to do.
Design for common circumstances first.
“ — Caroline Jarrett12
STYLING
By default, radio buttons (and checkboxes) are rendered
quite small. This makes them hard to click or tap, especially
for people with motor impairments.
We can increase the size using CSS, but this isn’t as simple
as it sounds. In “Making radio buttons and checkboxes eas-
ier to use,” Robin Whittleton explains that the way browsers
respond to CSS differs.13
Some browsers, such as Internet Explorer 8, won’t apply size

changes, but undesirable space will be created around the
radio button.
12 http://smashed.by/designmantra
13 http://smashed.by/govukissues
A Checkout Form 87
Other browsers, such as Firefox on OS X (10.10), will increase

the size but the radio buttons will appear blurry. Finally,
there are browsers that will apply the changes without
making them blurry.
Fortunately, a radio button’s label acts as a proxy for the

radio button itself. That is, when clicked, the radio button
will become checked (or unchecked, depending on state).
Unfortunately, many users don’t realize they can do this.14
This is hardly surprising because labels have very little to
signify that clicking them would do anything different to
regular copy.
To give users a better chance, we can color them gray and

make them respond to the mouse on hover. However, even
with these enhancements, some users may still be unaware.
GDS’s research showed this to be the case, which is why
they embarked on developing custom radio button controls.
The problem with creating custom controls is that you have

to reimplement all the behavior that is provided natively
for free. This is very involved, and despite GDS’s in-depth
attempts, they aren’t without their problems.15
14 http://smashed.by/govukcheckboxes
15 http://smashed.by/govukissues
88 Chapter 2
5. Delivery Notes
Imagine you’re at work. You receive a notification to say
your item is being delivered. When you arrive home, instead
of seeing the package, you find a card saying it couldn’t be
delivered because it was too big to fit through the letterbox.
Frustrating.
A delivery note, which you can provide at your discretion,

stops this from happening. The delivery note tells the
delivery person what to do if you’re not home. Perhaps you’d
prefer it to be left with a neighbor, or inside your recycling
bin, which Amazon refers to as a “safe place.” This, by the
way, works surprisingly well.
Delivery notes field with hint explaining what to enter.

A Checkout Form 89
<div class="field">
<label for="notes">
<span class="field-label">Delivery notes (optional)</span>
<span class="field-hint">Tell us where to leave your
package in case you’re not in. For example,
“Leave it with my neighbor”.</span>
</label>
<textarea id="notes" name="notes"></textarea>
</div>
The textarea is similar to a text box except it allows users

to enter multiple lines of text, which is particularly appro-
priate for a delivery note. (Remember: the size of the field
gives users a clue as to the length of content needed.)
While this question adds value, we need to understand how

it will be used by the delivery person as this may influence
the design. In this case, the viewport on the device is small
and can’t be scrolled, so we need to limit the amount of text
that can be entered.
LIMITING TEXT
Limiting the amount of text a user can type can and should
be handled by validation, as set out in chapter 1, “A Registra-
tion Form.” But there are some additional considerations.
The maxlength attribute (which takes a number value) lim-
its the amount of a text a user can type. As soon as the limit
is reached, the browser will ignore the input.
90 Chapter 2
Support for this attribute on a textarea is both lacking and

buggy.16 But even if it was well supported, it’s not recom-
mended because some users don’t look at the screen as
they type — they are focused solely on the keyboard. When
a user enters a lot of text, they’ll look up to find half their
entry has been truncated. Not good.
CHARACTER COUNTDOWN
Instead, we should let users type freely, and tell users how
many characters they have left. This way, users can see the
feedback when they finally look up at the screen and can
edit their entry in response. If they don’t notice the feed-
back, an error will
be shown when they
submit the form,
thanks to the valida-
tion routine (set out
in chapter 1, “A Regis-
tration Form”).

The character countdown
telling users how many
characters they have left.
16 http://smashed.by/maxlength
A Checkout Form 91
To create this component, we need to use JavaScript to inject

a status box below the field.
<div>You have 100 characters remaining.</div>
Then we need to listen to the textarea’s keydown event,

which is the event that fires as the user types:
function CharacterCountdown(input, options) {

this.input = $(input);
this.input.on("keydown", $.proxy(this, 'onKeydown'));
// ...
};
The event listener will then check the length of the typed
value against the configurable maxLength to calculate how
many characters are remaining. This is then injected into
the status box:
CharacterCountdown.prototype. {
var remaining = this.options.maxLength—this.field.val().
length;
this.status.html(this.options.message.replace(/%count%/,
remaining));
};
92 Chapter 2
Live Regions
The trouble is, this status is only determinable by sighted

users. To give screen reader users a comparable experience
(inclusive design principle 1), we should make sure this infor-
mation is communicated to them too.
Screen readers will normally only announce content when

it is focused, but live regions announce their content when
it changes. This means we can communicate to screen
reader users without asking them to leave their current
location. In this case, it means users can continue to type
into the textarea.
<div role="status" aria-live="polite">You have 100

characters remaining.</div>
Notes:
• The aria-live="polite" property17 and the status

role18 are equivalent. Both are provided to maximize
compatibility across platforms and screen readers (in
some setups, only one or the other is recognized).
17 http://smashed.by/aria
18 http://smashed.by/ariastatus
A Checkout Form 93
• The equivalent alert and assertive values mean the

current readout of the screen reader will be interrupted
to announce the live region’s new content. In this case,
interrupting the user as they’re typing is aggressive. So
we can keep status and polite values, which means
the contents are announced after the user stops typing
for a moment.
Announcing Only When It’s Critical
The character countdown we’ve designed so far has several

provisions that carefully help the user as they type their
entry. First, it doesn’t stop the user typing when they exceed
the maximum amount of characters. And second, it will
only inform users when they stop typing.
Even so, if the user is able to type a large amount of char-

acters, or if the maximum is rarely exceeded, it seems a bit
overbearing to interrupt users at all.
A more considerate approach might be to give feedback

when users get close to the limit. To do this we could add a
critical percentage option. Setting this to 10%, for example,
when the maximum is 100 would mean users start being
informed when they have typed 90 characters.
94 Chapter 2
6. Payment
It’s hardly surprising that most transactions are abandoned
at the payment page. Not only is this screen shown toward
the end of the journey (when users have had the most time
to reconsider their decision and used up a lot of energy), but
they may have to stop and find their credit card.
Fortunately, there are some usability provisions we can

apply here. By using autofill, removing unnecessary fields,
using the right input types and crafting label (and hint) text,
we can drastically reduce friction and keep users on-task.
REMOVING FIELDS
There are a number of details on a credit or debit card: name
on card, card number, valid from date, expiry date, issue
number, security number; all of these are commonly found
on payment forms. But not all of these details are needed to
process a payment.
When we designed Kidly’s checkout flow, chief technol-

ogy officer Øyvind Valland carefully picked Stripe19 as the
payment provider. This way, we didn’t have to worry about
PCI compliance20 and the cost of developing a solution from
scratch. Here’s the payment form we ended up with:
19 http://smashed.by/stripe
20 http://smashed.by/whatispci
A Checkout Form 95
You’ll notice the valid from

date, which is often provided
on a payment form is miss-
ing, so I spoke with Øyvind
to find out why. Here’s what
he said:
We don’t need to ask for “Valid
“ From.” Only a handful of debit

cards show those, and it provides
more hassle for the customer
to enter than benefit to us in
verifying card details. That is, if
the card is stolen, having to enter
a valid from date isn’t going to
Payment form with four fields: card
stop the thief. number, expiry date, security number, and
same as billing address checkbox (checked
by default).
He went on to talk about the
billing address, which is the
address to which the card is registered:
Only the numerics contained in card details are used for verification.
“ That is, the house number is used, but not street name. We ask for it
for our records. Being able to eyeball this stuff is handy in any situa-
tion where you have to query what’s happened. Besides, some people
expect that they’ll have to provide an address.
96 Chapter 2
Øyvind is not a designer per se, but his input into the design
process was crucial. Many of us assume that back-end
developers don’t care about the user experience, but tapping
into their knowledge is immensely valuable.
Design is a team sport, and so we should treat it as one. By

designing (and researching) with a diverse set of people,
we’ll frequently end up producing a far better experience.
We should also question the presence of form fields. If you
look at other people’s designs and assume something has to
be a certain way, we’ll never improve micropatterns such as
these. Proving assumptions are correct or otherwise is an
essential weapon in a designer’s arsenal.
When researching this section for the book, I decided to

speak to Stripe to see if we could reduce the amount of
fields even more. Here’s what they said:
The bare minimum information needed to attempt a valid payment
“ is card number, CVV, and expiration date. Additional information

will allow the card-issuing bank to make a more informed decision
about accepting or declining the payment, so while more infor-
mation isn’t required, it will improve your chances of the payment
succeeding. It will also provide you more information for verifying
the payment is valid and authorized, and therefore can help reduce
the likelihood of a dispute and help you with contesting the dispute if
it does occur.
A Checkout Form 97
AUTOFILL
Most modern browsers can automatically fill in form fields,
by way of the autocomplete attribute. When the user focuses
a particular field, the browser checks if it has that information
stored — if it does, the user can select it without having to type.
Chrome autofill: used 9 billion times/month; saves an average of 12
“ seconds; 1.25 million days saved/month

— Luke Wroblewski, October 24, 201721
Since iOS 8, Safari lets users scan their card using the
iPhone’s camera — it uses the same mechanism to automati-
cally fill out those fields.
Not only does this drastically reduce the amount of effort

to complete the form, but it also eliminates the chance of
typos: two very helpful improvements to a form that has the
highest drop-off rates in ecommerce.
As mentioned earlier, autofill is enabled with the autocomplete

attribute. Most modern browsers support it, but some older
browsers offer similar functionality by using the name attri-
bute instead. For the widest support, you should specify the
correct values for both attributes as shown below.22
21 http://smashed.by/chromeautofill
22 You can refer to the full list of available values in the HTML
specification: http://smashed.by/autocompletespec
98 Chapter 2
<div class="field">
<label for="ccname">
<span class="field-label">Name on card</span>
</label>
<input type="text" id="ccname" name="ccname" autocomplete="cc-name">
</div>
<div class="field">
<label for="cardnumber">
<span class="field-label">Card number</span>
</label>
<input type="text" id="cardnumber" name="cardnumber"
autocomplete="cc-number">
</div>
<div class="field">
<label for="expdate">
<span class="field-label">Expiry date</span>
</label>
<input type="text" id="expdate" name="expdate"
autocomplete="cc-exp">
</div>
<div class="field">
<label for="cvc">
<span class="field-label">Security code</span>
</label>
<input type="number" id="cvc" name="cvc" autocomplete="cc-csc">
</div>
<legend>
<span class="field-legend">Is your billing address the
same as delivery?</span>
</legend>
<div class="field-checkbox">
<label for="things">
<input type="checkbox" name="things" value="" id="things"
checked>Yes, it’s the same
</label>
</div>
</fieldset>
A Checkout Form 99
NUMBER INPUT
The number input (<input type="number">) lets mobile
users more quickly type a number via a numeric keypad. On
desktop, the input will contain increment and decrement
buttons called spinners, which make it easy to make small
adjustments without having to select and type.
You might think the number input is

appropriate for the card number, expiry
date, and CVC number — after all, they
all consist of numbers. But it’s a lot more
complicated than that. By looking at
Number input
what the spec says, what browsers do, with tiny spinner
and what users want, we can more easily buttons inside.
determine when the number input is
appropriate or not.
Let’s start with some definitions. Wikipedia says that:23
A number is a mathematical object used to count, measure, and label.
“ [...] numerals are often used for labels (as with telephone numbers),
for ordering (as with serial numbers), and for codes (as with ISBNs).
23 http://smashed.by/number
100 Chapter 2
Most of us think of numbers this way. We use them to

count and measure, but equally we use them in dates and
codes. However, the HTML specification only agrees in part
with this definition. It says that:24
The type=number state is not appropriate for input that happens
“ to only consist of numbers but isn’t strictly speaking a number. […]

When a spinbox interface is not appropriate, type=text is proba-
bly the right choice (possibly with a pattern attribute).
In other words, numbers and numerals are different. Num-

bers represent an amount of something such as:
• my age (announced “thirty-four years old”)

• the price of an apple (announced “forty-five pence”)
• the time it took me to cook breakfast (announced
“ten minutes”)
Conversely, numerals might be used for dates and codes

such as:
• birth date (announced “nineteenth of June,

nineteen eighty-three”)
• pin code (announced “eight, double five, three,
two, six”)
24 http://smashed.by/typenumber
A Checkout Form 101
There’s a difference between the way these values are

announced. Understanding this helps us see that while
the way browsers implement the number input may seem
buggy at first — it isn’t.
For example, IE11 and Chrome will ignore non-numeric

input such as a letter or a slash. Some older versions of iOS
will automatically convert “1000” to “1,000.” Safari 6 strips
out leading zeros. Each example seems undesirable, but
none of them stop users from entering true numbers.
Some numbers contain a decimal point such as a price;

other numbers are negative and need a minus sign. Unfor-
tunately, some browsers don’t provide buttons for these
symbols on the keypad. If that wasn’t enough, some desktop
versions of Firefox will round up huge numbers.
In these cases, it’s safer to use a regular text box to avoid

excluding users unnecessarily. Remember, users are still
able to type numbers this way — it’s just that the buttons
on the keypad are smaller. To further soften the blow, the
numeric keyboard can be triggered for iOS users by using
the pattern attribute.25
<input type="text" pattern="[0-9]*">
25 http://smashed.by/govuka11y
102 Chapter 2
In short, only use a number input if:
• incrementing and decrementing makes sense

• the number doesn’t have a leading zero
• the value doesn’t contain letters, slashes, minus
signs, and decimal points.
• the number isn’t very large
Let’s apply these rules to the expiry date. Incrementing it

doesn’t make sense, the number could start with a zero,
and credit cards put a slash in the middle of the expiry date
which users should be able to copy. Using a number input is
not only inappropriate, but it creates a jarring experience as
the user types a slash which would be ignored.
We’ll look at appropriate use cases of the number input in

the next chapter.
A Note about the Telephone Input
The telephone input (<input type="tel">) is sometimes

used as a makeshift number input because it gives users
the benefits of a number-specific keypad on mobile without
some of the pitfalls discussed earlier.
A Checkout Form 103
Apart from semantic incorrectness, I’ve not come across

practical reasons not to use it. However, ignoring the stan-
dards specification is generally not recommended because
we don’t know how browsers and devices might handle it in
the future. For example, perhaps a browser’s autocomplete
routine will try to fill in the user’s telephone number instead
of the credit card number. Or a device may choose to load
the user’s contact address book, from which they can choose
an appropriate number.
Both of these features would be suited to a legitimate

telephone field, but not for anything else that happens
to consist of numbers. Updating your implementation in
response to a new browser exposing new behavior is not an
ideal strategy.
FORGIVING BAD INPUT

In “A Registration Form” we briefly talked about forgiving
little input mistakes. In fact, the success of the internet is
largely down to its robustness, thanks to what’s known as
Postel’s law:
“ Be conservative in what you send; be liberal in what you accept.

104 Chapter 2
We can apply this principle to the fields in the payment

form. For example, a card number typically appears as six-
teen digits split into four parts by spaces. Some users may
type the space; others might not.
Similarly, for the expiry date some users might type a slash,
others may leave it out. Whether it’s a slash or space, or a card
number, or an expiry date, we should be forgiving by strip-
ping whitespace and normalizing the format where possible.
CARD VERIFICATION CODE (CVC) FIELD

Every payment provider needs the user’s CVC number, usu-
ally the last three digits found on the back of the card.
The first problem is that sites don’t always refer to this field
as the CVC number. Sometimes it’s referred to as a security
code number or card verification value (CVV). Being speci-
fied as an acronym doesn’t help either. And to top it off, on
the card the number is never accompanied by a description,
making it hard to reconcile the requirements.
To fix this, we should employ the hint text pattern to tell

users exactly what it is and where to find it. For example:
“This is the last three digits on the back of the card.”
A Checkout Form 105
BILLING ADDRESS TOGGLER

The billing address is the address to which the card is reg-
istered and is needed to process a payment. For most users,
their billing address is the same as the delivery address. As
the user has already provided this information, we can use
it to improve the experience.
First, we need to add a checkbox field which asks the user if

their billing address is the same as their delivery address.
This way, users only have to fill out the billing address on
the rare occasion that it’s different. As it’s the most common
scenario, it’s checked by default.
We can enhance the experience by hiding the billing

address until the user unchecks the checkbox. This is a form
of progressive disclosure, which means showing informa-
tion only when it becomes relevant.
Here’s the complete script with notes to follow.
function CheckboxCollapser(checkbox, toggleElement) {

this.checkbox = checkbox;
this.toggleElement = toggleElement;
this.check();
this.checkbox.on('click', $.proxy(this,
'onCheckboxClick'));
};
106 Chapter 2
CheckboxCollapser.prototype. {
this.check();
};
CheckboxCollapser.prototype.check = function() {
if(this.checkbox.prop('checked')) {
this.toggleElement.addClass('hidden');
} else {
this.toggleElement.removeClass('hidden');
}
};
Notes
• The constructor function takes the checkbox element

and toggle element (the billing address container) as
parameters and assigns them to this so they can be
referenced in the other methods.
• The state of a checkbox is persisted on the server. As
it could be checked or unchecked the state is checked
on initialization and hides or shows the billing address
fields accordingly.
• The same routine is performed each time the checkbox
is clicked. The check() method checks to see if the
checkbox is checked or not. If so, the billing address
container is hidden, otherwise it’s shown, by adding
and removing the hidden class.
A Checkout Form 107
7. Check Your Answers Page

Even though we’ve collected all the information required
to process the order, we should first let users review their
order. As counterintuitive as this may sound, adding an
extra step in the flow actually reduces effort and likely
increases conversion.
Take Jack (I made him up), a father of two infants. It’s the
middle of the night, and his newborn baby is crying incon-
solably. Naturally, Jack’s tired and stressed. To make things
worse, there are no more nappies.
He grabs the phone, adds nappies to the basket, fills out all
the details and submits the order. Great — except it isn’t. He
ordered the wrong size nappies and paid with the wrong
card. What Jack entered was valid, but still a mistake.
We can save Jack a lot of frustration by giving him the chance

to review the order on a separate page. That way he can focus
on the order details. Remember, filling out forms and check-
ing information are two different mental contexts.
This also saves your (client’s) business time and money. If

Jack wants to cancel the order, then handling calls and pro-
cessing would be costly — especially if the business offers
free returns.
108 Chapter 2
This shows that relying solely on completion time as a

metric for success is dangerous. You should also look at how
accurate people’s orders are by checking how often items
are returned.
As this is the final step in the flow, the button’s text should
be set to “Place order” or similar. Leaving it as “Continue”
would mislead the user into thinking there is another step
to complete, which is likely to result in more cancelled
orders and strain on the customer service department. This
would increase operating costs and shows that good design
is also good for business.
MAKING CHANGES
Every piece of information gathered during checkout
should be represented on the review page. Users shouldn’t
have to go back to check information — that would defeat
the purpose of the page. Users should only need to go back
if they spot a mistake.
Users can click Edit to make amendments, which is another

advantage of using the one thing per page pattern. As pages
are small they will load fast; as each page has just one thing
on it, making a change is simple.
A Checkout Form 109
Left: the review page with edit links. Right: the user editing their mobile num-
ber, having spotted a mistake.
When the user makes a change, they are taken back to the
‘check your answers’ page again for another review, which
puts users firmly in control and reduces stress and anxiety.
8. Confirmation Page
Confirmation pages are so much more than just confirming
the order. Neglecting the user experience here is a great way
to lose out on future business.
We’ve all probably experienced neglect after purchasing

goods. For example, if you want to take out insurance, you
110 Chapter 2
call the free sales number and are quickly put through to
a helpful agent. Parting with money is usually made easy.
But when you need to make a claim, it’s more painful: the
number isn’t free and calls take a long time be answered. All
very stressful.
A confirmation page is the first opportunity to start forging a

long-term relationship. And this is done in two ways: by look-
ing after the user, and giving them an incentive to come back.
You should tell users what happens next, such as when deliv-
ery will take place, and what to do if something goes wrong.
It’s also the best time to ask users to sign up (if they checked
out anonymously). As we have most of the information
to hand, we only need to ask for a password, making this
step both optional and easy. Users should have had a good
experience up to now, which should naturally encourage
sign-up.
By giving users value, they’ll hopefully want to sign up.

By value, I mean something as simple as offering a faster
checkout next time, or offering them a discount on their
next order. Depending on the service, you might even ask
users to tweet or Instagram their purchase in return for
a voucher. Whatever it is, now’s as good a time as any to
mention it.
A Checkout Form 111
Up to now, we’ve also been sure to use plain and simple

language for labels, hints, and errors. On the confirmation
page, there is an opportunity to let your brand’s personality
shine through because the important stuff has already been
done. MailChimp’s confirmation page shows their mascot
Freddie giving you a virtual high five. Nice.
MailChimp’s confirmation screen has Freddie high-fiving.
Here’s a checklist of things to consider including on the

confirmation page:
• a reference number
• contact details
• what happens next and when
• what to do if something goes wrong
• ask users to sign up in return for something
112 Chapter 2
• ask users to spread the word to get a voucher

• links to further information, if they might be useful
• a link to your feedback page
The Second-Time Experience

I’ve made so many purchases with Amazon that I can’t even
remember Amazon’s first-time experience. That is to say,
I’ve made hundreds of purchases as a second-time user, and
only one as a first-time user.
Once everything is good with the first-time experience, it’s

time to focus on the second-time experience. As we have
collected and stored all the user’s information, we shouldn’t
have to ask for it again.
We can bypass most of the steps by sending users straight

to the ‘check your answers’ page. This way, users get a
reminder of their default preferences with the chance to
make amends should they wish to.
All in all, this significantly reduces the effort and improves

conversion. Remember Jack from earlier: the next time he’s
out of nappies, buying them should be a breeze, something
I’m sure he’d appreciate.
A Checkout Form 113
Layout
Up to now, we’ve focused on the design of the form within
each page, but we haven’t considered the interface holis-
tically. In fact, this is one of the dangers of composing
interfaces out of predefined smaller components: the overall
design can end up neglected.
Usually, checkout pages are given a special and more

streamlined layout that helps reduce noise and keep users
on-task. For example, the header usually contains a logo,
security note, and accepted cards.
By omitting navigation and search, users can focus on

checking out, which speaks to inclusive design principle 6,
“Prioritize content.”
INDICATING PROGRESS
Progress bars or indicators are often used within check-
out because — at least in theory — they give users an idea
of where they are and how long’s left. Despite the sound
reasoning, there isn’t much evidence to show that progress
bars are all that useful or even noticed. For example, the UK
government’s Carer’s Allowance team removed a 12-step
progress bar with no effect on completion rates or times.26
26 http://smashed.by/sharedspaces
114 Chapter 2
A progress bar showing the user they are at step 4.
Progress bars pose some practical design challenges too.

First, they take up a lot of space at the top of the page, which
is particularly important on mobile where they can push the
main content down. Second, fitting an accessible progress
bar with clear labeling into a small viewport is nigh on
impossible.
If that weren’t enough, they’re even trickier to design when

the journey consists of conditional steps. Imagine a check-
out that offers collection or delivery. If the user chooses
to collect, they’re taken down a different path where they
won’t need to give their payment details (as they’ll pay
in-store).
The problem is that a progress bar should tell the user what
steps exist in advance. But the steps are based on users’
answers. Either you show every possible step, which is
misleading, or you update the progress bar (by removing
or adding steps) as you go, which somewhat defeats the
purpose of having one.
A Checkout Form 115
Misleading progress bar because the payment step isn’t applicable when
collecting in-store.
Not including a progress bar prioritizes the main content

by moving it further up the page, which speaks to inclusive
design principle 6 (again), “Prioritize content.” In any case,
having meticulously designed the journey to be as simple as
possible, users should get to the end quickly, which reduces
the need for the progress bar.
For these reasons, consider starting without a progress bar.

Then test your journey to see if users struggle. Remem-
ber, it’s far easier (and cheaper) to add features than it is to
remove them later on.
Note: Some forms are especially long — a lot longer

than a checkout flow. In this case, you might need some
indication of progress, which is something we’ll look at in
chapter 10, “A Really Long and Complicated Form.”)
116 Chapter 2
Progress Step Text
If you decide to give users an indication of progress, first try

adding the step number inside the heading:
<h1>Payment (Step 3 of 4)</h1>
Visual Progress Bar
If research shows that a more prominent progress bar is

useful, then you can include one — but there are a few
things to consider.
First, keep the text inside the <h1> so screen readers get a
comparable experience (inclusive design principle 1). Second,
you’ll need to hide it from sighted users like this:
<h1>Payment <span class="visually-hidden">Step 3 of 4

</span></h1>
The visually-hidden class contains a special set of proper-

ties that hide the element visually, while still ensuring it’s per-
ceivable to screen reader users when the <h1> is announced.
.visually-hidden {
border: 0!important;
clip: rect(0 0 0 0)!important;
height: 1px!important;
margin: -1px!important;
A Checkout Form 117
overflow: hidden!important;
padding: 0!important;
position: absolute!important;
width: 1px!important;
}
As the visual progress bar is redundant for screen reader

users, we can use the aria-hidden="true" attribute, which
stops it being announced:


<div aria-hidden="true">
Visual progress bar here
</div>
Order Summary
When you’re shopping in a physical shop, you pick up items
and place them in your shopping basket. Eventually, you
checkout at the till. All the while, you can see what you’re
buying. Sometimes, at the last minute, you change your
mind and take an item off the conveyor belt. Or you realize
you forgot something and dash off to get it.
“ The system should always keep users informed about what is going
on, through appropriate feedback.
— Jakob Nielsen, “10 Usability Heuristics for User Interface Design”27
27 https://smashed.by/usabilityheuristics
118 Chapter 2
Giving users a comparable experience digitally is import-

ant. We can place an order summary panel on every page
to keep users informed, without them having to remember
what they’re buying, freeing up their mental energy to focus
on checking out.
The order summary panel gets populated as the user completes each screen.
The email screen containing just the basket.
The mobile number screen, now containing the email address previously
populated.
A Checkout Form 119
As the user completes each step, the order summary will

be populated with more information. For example, on the
email address screen (step 1), the summary only shows what
they’re buying. On the mobile number screen (step 2), the
summary will also be populated with their email address,
and so forth. If the user spots a mistake, they can jump back
to any previous step by clicking the Edit link — just like the
‘check your answers’ page.
While the summary panel is important, it’s less important

than the form. So it should be placed beside the form on
desktop, where there’s enough space to do so, and below the
form on mobile, where there’s not enough room.
BACK LINKS
As the user is following a linear flow, we need to consider
the need to step back. The browser’s back button provides
this functionality for free, but some people mistrust it
because of bad past experiences when their data was lost.
Ajax-driven and modal-heavy sites haven’t helped matters,

because clicking the browser’s back button often goes
against user’s expectations.28 Thankfully, the checkout
doesn’t need Ajax or modal dialogs so this isn’t a problem,
but we still need to ensure the browser’s back button works
28 http://smashed.by/backbutton
120 Chapter 2
as expected. Users expect it to take them to the previous

page in the state they left it.
Research might show you that it’s useful to include a back

link within the interface itself, and that users will be more
inclined to trust it. In this case, position the link at the top left
of the page. By placing it at the top of the page, users can see
that they can go back if they need to. And they’re less likely to
fill out the form before hitting back and losing their data.
The back link positioned at top-left of the form.

A Checkout Form 121
Summary
In this chapter, we started out by looking at the one thing
per page pattern, which helps to break down large forms
into small chunks, making it easy for users to fill out and
make amendments.
We then looked at capturing optional information, making

choices with radio buttons, entering long-form content via
the textarea and several ways to improve the payment
form experience.
After that, we looked at other issues: giving users the ability

to review their order, improving the experience for second-
time users, and considering the overall design.
Demos
• Optional telephone field:
http://smashed.by/telinputdemo
• Delivery radio buttons:
http://smashed.by/deliveryoptionsdemo
• Delivery notes with countdown:
http://smashed.by/charcountdowndemo
• Payment form: http://smashed.by/paymentformdemo
122 Chapter 2
CHECKLIST
• Ask questions in a sensible order.

• Let users checkout anonymously.
• The width of the field should match the required input
when the length is known.
• Use a fieldset and legend to give radio button and
checkbox groups an accessible label.
• Add extra questions if they add value. Remember, com-
pletion time is not the only useful metric for success.
• Let users check their answers before submission.
• The confirmation page is the end of the transaction but
the start of the relationship.
• Store people’s information to improve the second-time
experience and increase conversion.
• Don’t break the back button.
A Flight Booking Form 123
A Flight Booking Form
I
n this chapter, we’ll design a flight booking service. At
first glance this may seem a bit niche, especially when
compared to “A Registration Form” and “A Checkout
Form.” However, we’re going to explore several complex
problems that, in the end, will result in reusable patterns
— patterns that are very much transferable to other prob-
lem domains, such as booking a cinema ticket, or even a
hotel room.
Booking a flight consists of many discrete steps. The first

few steps simply collect the user’s preferences: where to fly,
when to fly, and who’s flying. Once we have that, we can
give users a choice of available flights, followed by choosing
where to sit. Finally, users will have to make payment (we
covered payment patterns in the previous chapter).
1. Where to Fly
First, users have to choose an origin and destination; that
is, places to fly from and to. Without this information, the
service can’t offer any flights. What’s the best way of asking
users for this information?
124 Chapter 3
We should try to use the features that are native to the browser.
They are familiar (by convention) and fully accessible out of
the box. They also require far less work to implement.
You’d be forgiven for thinking you were spoiled for choice

when it comes to form controls: select boxes, radio buttons,
text boxes and, more recently, datalists. The choice is yours
— except it isn’t. Not all of these options are suitable. Let’s
look at some of the pros and cons of each.
SELECT BOX
Also known as dropdown menus, select boxes hide options
behind a menu. Clicking the select box reveals the options.
Once an option is selected, the menu collapses. Select
boxes are often used for their space-saving qualities.
What’s most interesting, though, is why we need to save
space in the first place.
Destination field as a select box. Left: collapsed. Right: expanded with options
showing.
Often an interface is crammed with features, usually to

please stakeholders, not users. It’s understandable, then,
that learning ways to hide discrete pieces of an interface has
become part of a designer’s skill set. But design is about so
much more than saving space. After all, if an interface really
is crammed, then our first job as designers is to declutter it.
In her talk “Burn Your Select Tags,” Alice Bartlett shares the
user research she undertook at the UK’s Government Digital
Service (GDS).1
In short, select boxes are hard to use. Besides hiding options

behind an unnecessary extra click, users generally don’t
understand how they work. Some users try to type into
them, some confuse focused options with selected ones.
And, if that weren’t enough, users can’t pinch-zoom the
options on certain devices.
Usability expert, Luke Wroblewksi wrote an article called

“Dropdowns Should be the UI of Last Resort.”2 In it, he sug-
gests some better alternatives, some of which we’ll discuss
later in this chapter.
1 http://smashed.by/burnselecttags
2 http://smashed.by/dropdownlastresort
126 Chapter 3
Destination field using radio

buttons.
RADIO BUTTONS
Radio buttons, unlike select boxes, are generally well under-
stood and easy to use, not least because they don’t hide
options. They are exposed, making them easy to compare,
scan, and select. They’re also malleable; that is, they let us
use whatever content, in whatever format we want, inside
the related label (more on that shortly).
The problem with radio buttons is that they’re less suitable

when there are many options. An airline could fly to hun-
dreds of destinations, making the page long and hard to
scan. This in turn means users have to scroll a lot more.
Don’t get me wrong, users are more than happy to scroll,

and we shouldn’t use this as a crutch for changing course.3
But if we can eliminate the need to scroll without introduc-
ing new problems, we should.
3 http://smashed.by/ppldontscrollmyth
SEARCH INPUT
A search box (<input type="search">) is similar to a regu-
lar text box (<input type="text">). A search box, however,
lets users clear the field by tapping a delete icon, or pressing
Escape when focused. With a text box, you have to select
the text and press Delete, which takes a little longer.
Destination field as a search input.
Using a search box is useful when searching a large amount

of dynamic data, such as searching Amazon’s product
catalog. Airlines, however, fly to a finite set of destinations
known in advance of the user searching. Letting users
search unassisted like this could easily result in a page with
no results, due to typos or a data mismatch.
DATALIST
Users need a control that lets them filter a long list of
destinations. A control that marries the flexibility of a text
box with the assurance of a select box. This type of con-
trol goes by many different names, including: type ahead,
predictive search, and combo box; but we’ll refer to it as an
autocomplete control.
128 Chapter 3
Autocomplete controls work by filtering options (destina-

tions in this case) as the user types. As suggestions appear,
users can select one quickly, automatically completing the
field. This saves users having to scroll (unless they want to),
while also being able to forgive small typos.
HTML5’s <datalist> combines with a text box (<input

type="text">) to create a native autocomplete control,
which is unfortunately too buggy for use on the open web.4
However, if your project is locked down to a few known
browsers that don’t have these bugs, then a native solution
might work for you.
Destination field as a datalist.
But, having already defined our design principles in the

introduction, we know we want to design an inclusive expe-
rience — one that works for as many people as possible, no
matter their choice of browser or mobile device.
4 http://smashed.by/datalistcaniuse
AN AUTOCOMPLETE CONTROL
By creating a custom autocomplete component from
scratch, there’s an opportunity to create a powerful expe-
rience that also allows for common typos and endonyms
(discussed later). A word of warning though: we’re going to
break new ground. Designing a robust and fully inclusive
autocomplete control is hard work — but that’s what our job
is all about.
Do the hard work to make it simple.
“ — GDS Design Principle 4
Accessibility expert Steve Faulkner has what he calls a punch

list, which is a set of rules to make sure that any custom Java-
Script component is designed and built to a good standard.5
The rules state that a component should:
• work without JavaScript

• be focusable with the keyboard
• be operable with the keyboard
• work with assistive devices
5 http://smashed.by/webcomponents
130 Chapter 3
The Basic Markup
To satisfy the first rule we need to make sure the interface

works in the absence of JavaScript. This means starting with
a native form control that browsers provide for free.
Having already discussed the options above, we know that

there are too many options for radio buttons; a search box
requires an unnecessary round-trip to the server, and can
lead to zero results; and the datalist is too buggy. By process
of elimination, we’re left with a select box.
<div class="field">
<label for="destination">
<span class="field-label">Destination</span>
</label>
<select name="destination" id="destination">
<option value="">Select</option>
<option value="1">France</option>
<option value="2">Germany</option>
<option value="3">Spain</option>
</select>
</div>
The Enhanced Markup
When JavaScript is available, the Autocomplete() construc-

tor function will enhance the basic HTML to look like this:
<div class="field">
<label for="destination">
<span class="field-label">Destination</span>
</label>
<select name="destination" aria-hidden="true"
tabindex="-1" class="visually-hidden">

</select>
<div class="autocomplete">
<input aria-owns="autocomplete-options--destination"
autocapitalize="none" type="text" autocomplete="off"
aria-autocomplete="list" role="combobox" id="destination"
aria-expanded="false">
<svg focusable="false" version="1.1" xmlns="http://www.
w3.org/2000/svg">

</svg>
<ul id="autocomplete-options--destination"
role="listbox" class="hidden">
<li role="option" tabindex="-1" aria-selected="false"
data-option-value="1" id="autocomplete_1">
France
</li>
<li role="option" tabindex="-1" aria-selected="true"
data-option-value="2" id="autocomplete_2">
Germany
</li>

</ul>
<div aria-live="polite" role="status" class="visually-hidden">
13 results available.
</div>
</div>
</div>
132 Chapter 3
Select Box and Text Box Notes
Even though users will no longer interact with the select

box, we can’t remove it completely. If we were to remove
the select box from the document (or hide it with display:
none;) then its value wouldn’t be sent to the server on sub-
mission. This is important because the text box value differs
from the select box value that will be submitted.
Hiding the select box while still having its value submitted
involves a number of techniques in combination. The vis-
ually-hidden class and aria-hidden="true" attribute (as
first set out in “A Checkout Form”) hide the select box from
sighted and screen reader users respectively. The tabin-
dex="-1" attribute stops keyboard users from being able to
focus it.
Note that the select box id attribute is transferred over to

the text box because the label must be associated to it so it’s
read out in screen readers, and to increase the hit area of the
control (as explained in chapter 1). The select box, however,
no longer needs an id — it has effectively become a hidden
input. Conversely, the name attribute isn’t needed on the text
box because its value isn’t sent to the server — it’s purely for
interaction purposes and is used as a proxy to set the select
box value.
The role="combobox" attribute means the control is

announced as a combo box instead. A combo box, according
to MDN, is “an edit control with an associated list box that
provides a set of predefined choices.”6 The aria-autocom-
plete="list" attribute tells users that a list of options will
appear. The aria-expanded attribute tells users whether
the menu is expanded or collapsed by toggling its value
between true and false.
The autocomplete="off" attribute stops browsers from

showing their own suggestions, which would interfere with
those offered by the component itself. Finally, the autocap-
italize="none" attribute stops browsers from automati-
cally capitalizing the first letter, something we’ll look at in
detail in the next chapter.
The SVG icon is layered on top of the text box using CSS.
Note the focusable="false" attribute, which fixes the
issue that in Internet Explorer SVG elements are focusable
by default.
Menu Notes
The role="list" attribute is used to communicate the

menu as a list, because it will be populated with a list of
options. Each option has a role="option" attribute.
6 http://smashed.by/combobox
134 Chapter 3
The aria-selected="true" attribute tells users which

option within the list is selected or not by toggling the value
between true and false.
The tabindex="-1" attribute means focus can be set to the

option programmatically when users press certain keys.
We’ll look at keyboard interaction later.
Finally, the data-option-value attribute stores the select

box option value. When the user clicks an autocomplete option,
the select box value is updated accordingly to keep them in
sync. This is what ties the enhanced interface (what the user
sees) with the select box (what the user can’t see) that’s used
to communicate to the server when the form is submitted.
Live Region
Sighted users will see the suggestions appear in the menu

as they type, but the act of populating the menu isn’t deter-
minable to screen reader users without leaving the text box
to explore the menu.
To provide a comparable experience (inclusive design principle

1), we can use a live region as first laid out in “A Checkout
Form” (see page 92). As the menu is populated, we’ll also
populate the live region with how many results are available;
for example, “13 results available.” With this information at
hand, users can continue typing to narrow the results fur-

ther, or they can select a suggestion from the menu.
As the feedback is only useful to screen reader users, it’s

hidden with the visually-hidden class again.
Typing into the Text Box
When the user types into the text box, we need to listen out
for certain keys by using JavaScript.
Autocomplete.prototype.createTextBox = function() {
// ...
this.textBox.on('keyup', $.proxy(this, 'onTextBoxKeyUp'));
// ...
};
Autocomplete.prototype. {
switch (e.keyCode) {
case this.keys.esc:
case this.keys.up:
case this.keys.left:
case this.keys.right:
case this.keys.space:
case this.keys.enter:
case this.keys.tab:
case this.keys.shift:
// ignore these keys otherwise the menu will show briefly
break;
case this.keys.down:
this.onTextBoxDownPressed(e);
break;
default:
this.onTextBoxType(e);
}
};
136 Chapter 3
The this.keys object is a collection of key codes (numbers)

that correspond to particular keys by their names. This is to
avoid magic numbers, which makes the code easy to under-
stand at a glance.7
The switch statement filters out the Escape, Up, Left,

Right, Space, Enter, Tab, and Shift keys. If we didn’t, the
default case would run and incorrectly show the menu.
Instead of filtering out the keys we aren’t interested in
responding to, we could have specified the keys that we are
interested in. But this would mean specifying a huge range
of keys, which would increase the chance of one being
missed, creating a broken experience.
We’re mainly interested in the last two statements; that is,

when the user presses Down, or the default case above, which
means everything else (a character, number, symbol, and so
on). In this case the onTextBoxType() function will be called.
// only show options if user typed something
if(this.textBox.val().trim().length > 0) {
// get options based on value
var options = this.getOptions(this.textBox.val().
trim().toLowerCase());
// build the menu based on the options
this.buildMenu(options);
7 http://smashed.by/magicnumber
// show the menu

this.showMenu();
// update the live region
this.updateStatus(options.length);
}
// update the select box value which
// the server uses to process the data
this.updateSelectBox();
};
The getOptions() method filters the options based on what

the user typed. We’ll look at the the filter function later.
Controls Should Have a Single Tab Stop
The autocomplete control is what’s known as a composite.

That just means it’s made up of several different interactive
and focusable parts — in this case, the text box and menu.
What’s important is that composite components should
have one tab stop. Here’s what the WAI-ARIA Authoring
Practices 1.1 specification has to say on the subject:8
A primary keyboard navigation convention common across all
“ platforms is that the tab and shift+tab keys move focus from one UI
component to another while other keys, primarily the arrow keys,
move focus inside of components that include multiple focusable
elements. The path that the focus follows when pressing the tab key
is known as the tab sequence or tab ring.
8 http://smashed.by/generalnav
138 Chapter 3
A radio button group, for example, is a composite control

that has one tab stop. Once the first radio button is focused,
users can use the arrow keys to move between the options.
Pressing Tab at anytime from within the group moves focus
to the next focusable control in the tab sequence.
The text box within our autocomplete control is naturally

focusable by the Tab key. Once focused, the user will be able
to press the arrow keys to traverse the menu, which we’ll
look at shortly. Pressing Tab when the text box or menu
option is focused should hide the menu to stop it from
obscuring the content beneath when not in use. We’ll look
at how to do this next.
Note: One way to make a composite control have just a

single tab stop is to use the aria-activedescendant
attribute.9 It works by keeping focus on the component’s
container at all times, but referencing the currently
active element. This doesn’t work for the autocomplete
component because the text box is a sibling (not a parent)
of the menu.
9 http://smashed.by/activedescendant
Hiding the Menu on Blur Is Problematic
The onblur event is triggered when the user leaves an

in-focus element. In the case of the autocomplete, we could
listen to this event on the text box. The virtue of using
the onblur event is that it will be triggered when the user
leaves the field by pressing Tab or by clicking or tapping
outside the element.
this.textBox.on('blur', function(e) {
// hide menu
});
The problem with this approach is that the act of moving

focus to the menu (even programmatically like we will)
triggers the blur event, which subsequently hides the menu.
This would make the menu inaccessible with the keyboard.
One workaround involves using the setTimeout() func-

tion, which allows us to put a delay on the event. In turn, the
delay gives us time to cancel the event (using clearTime-
out()) should the user move focus to the menu within that
time. This would stop the menu being hidden, making it
accessible again.
140 Chapter 3
this.textBox.on('blur', $.proxy(function(e) {
// set a delay before hiding the menu
this.timeout = window.setTimeout(function() {
// hide menu
}, 100);
}, this));
this.menu.on('focus', $.proxy(function(e) {
// cancel the hiding of the menu
window.clearTimeout(this.timeout);
}, this));
Unfortunately, there’s a problem with the blur event in

iOS 10. It incorrectly triggers the blur event on the text box
when the user hides the on-screen keyboard. This stops
users from accessing the menu altogether. There’s another
solution which we’ll look at next.
Listening to the Tab Key
Instead of hiding the menu on blur, we can use the keydown

event to listen out for when the user presses the Tab key.
this.textBox.on('keydown', $.proxy(function(e) {
case this.keys.tab:
// hide menu
break;
}
}, this));
Unlike the blur event, this approach doesn’t cover the case
where users blur the control by clicking outside of it. We
have to handle this case manually by listening to the doc-
ument’s click event, but being careful to work out what’s
clicked — we don’t want to hide the menu if the user clicks
within the control.
$(document).on('click', $.proxy(function(e) {
if(!$.contains(this.container[0], e.target)) {
// hide the menu
}
}, this));
The event handler is using jQuery’s contains() method,

which checks to see if what the user clicked (e.target) falls
outside the container (this.container[0]). If it’s outside,
the menu is hidden. The [0] is used because the contains()
method takes element nodes, not jQuery objects.
Moving to the Menu (Pressing Down)
When the text box is focused, pressing the Down key trig-
gers onTextBoxDownPressed() like this:
var option;
var options;
var value = this.textBox.val().trim();
142 Chapter 3
/*
When the value is empty or if it exactly
matches an option show the entire menu
*/
if(value.length === 0 || this.isExactMatch(value)) {
// get options based on the value
options = this.getAllOptions();
// show the menu
this.showMenu();
// retrieve the first option in the menu
option = this.getFirstOption();
// highlight the first option
this.highlightOption(option);
/*
When there’s a value that doesn’t have
an exact match show the matching options
*/
} else {
// get options based on the value
options = this.getOptions(value);
// if there are options
if(options.length > 0) {
// show the menu
this.showMenu();
// retrieve the first option in the menu
option = this.getFirstOption();
// highlight the first option
this.highlightOption(option);
}
}
};
If the user presses Down without having typed anything,

the menu will populate with every available option, and
focus to the first menu option. The same thing will happen
if the user types an exact match; this should be rare because
most users who notice the suggestion will select it — it’s
quicker that way.
The else condition will populate the menu with options that
match (if any), and again will focus the first menu option. At
the end of both scenarios the highlightOption() method
is called, which we’ll look at later.
Scrolling the Menu
As mentioned earlier, the menu may contain hundreds of

options. To ensure the menu stays visible within the view-
port, we need to use CSS.
.autocomplete [role=listbox] {
max-height: 12em;
overflow-y: scroll;
-webkit-overflow-scrolling: touch;
}
The max-height property works by letting the menu grow

up to a maximum height of 12em. Once the content inside
the menu surpasses that height, users can scroll the menu
thanks to the overflow-y: scroll property.
144 Chapter 3
The last property is non-standard and is used to enable

momentum scrolling on iOS. This ensures the autocomplete
control scrolls the same way as it would everywhere else.
Clicking an Option
Clicking or tapping a menu option should perform a num-

ber of discrete tasks, but before we get to them, let’s discuss
how we might listen to the click event.
The most basic approach involves adding a click event to

each of the options individually. But this is problematic for
two reasons.
First, each added event must be stored in memory. As there

are hundreds of option, they’ll use a lot of memory which
may impact performance. Second, the menu options are con-
stantly being updated as the user types. This means events
need to be constantly added and removed, which is computa-
tionally intensive and bothersome to manage with code.
Instead, we can use event delegation, which is made possi-

ble by the concept of event bubbling.10 Events originating
from lower down the document tree propagate (bubble up)
to the parent container, all the way up to the document root.
10 http://smashed.by/eventdelegation
In our particular case, we can add a single event listener

to the menu’s container and filter out all events that don’t
match the option elements we’re interested in. To do this,
we can use jQuery’s on() method, which has event delega-
tion built in.
Autocomplete.prototype.createMenu = function() {
//...
this.menu.on('click', '[role=option]', $.proxy(this,
'onOptionClick'));
//...
};
The click event is bound to the container (this.menu), but

will only trigger the event handler (onOptionClick())
when the event originated on an element with
role="option".
var option = $(e.currentTarget);
this.selectOption(option);
};
The event handler retrieves the option (e.currentTarget)

and hands it off to the selectOption() method. Normally,
we’d reference e.target, but as we’re using event delega-
tion, e.target would return the delegate (this.menu) which
146 Chapter 3
isn’t helpful. Whenever you’re using event delegation, you’ll

almost definitely be interested in the originating element
(e.currentTarget).
Autocomplete.prototype.selectOption = function(option) {
var value = option.attr('data-option-value');
this.setValue(value);
this.hideMenu();
this.focusTextBox();
};
The selectOption() function takes the option to be

selected and extracts the data-option-value attribute.
That value is passed to the setValue() method which pop-
ulates the text box and hidden select box. Finally, the menu
is hidden and the text is focused.
This same routine is performed when the user selects an

option with the Space or Enter keys. We’ll look at the menu
interactions next.
Menu Keyboard Interaction
Once focus is within the menu (by pressing Down while

the text box is focused), we need to let users traverse the
menu with the keyboard. To do this, we’ll listen to the
keydown event.
Autocomplete.prototype.createMenu = function() {
this.menu.on('keydown', $.proxy(this, 'onMenuKeyDown'));
};
case this.keys.up:
// Do stuff
break;
// Do stuff
break;
// Do stuff
break;
// Do stuff
break;
case this.keys.esc:
// Do stuff
break;
case this.keys.tab:
// Do stuff
break;
default:
this.textBox.focus();
}
};
148 Chapter 3
Key Action
If the first option is focused, set focus to the text
Up
box; otherwise set focus to the previous option.
Focus the next menu option. If it’s the last menu
Down
option, do nothing.
Tab Hide the menu.
Enter or Select the currently selected option and focus the
Space text box.
Escape Hide the menu and set focus to the text box.
Any other
Focus the text box so users can continue typing.
character
The Highlight Function
As noted above, the user can focus an option by pressing

the Up or Down keys. When this happens, the
highlightOption() method is called.
Autocomplete.prototype.highlightOption = function(option) {
// if there’s a currently selected option
if(this.activeOptionId) {
// get the option
var activeOption = this.getOptionById(this.
activeOptionId);
// unselect the option
activeOption.attr('aria-selected', 'false');
}
// set new option to selected
option.attr('aria-selected', 'true');
// If the option isn’t visible within the menu

if(!this.isElementVisible(option.parent(), option)) {
// make it visible by setting its position inside the
menu
option.parent().scrollTop(option.parent().scrollTop() +
option.position().top);
}
// store new option for next time
this.activeOptionId = option[0].id;
// focus the option
option.focus();
};
The method performs a number of discrete steps. First, it

checks to see if there’s a previously active option. If so, the
aria-selected attribute is set to false, which ensures the
state is communicated to screen reader users. Second, the
new option’s aria-selected attribute is set to true.
As the menu has a fixed height, there’s a chance that the

newly focused option is out of the menu’s visible area. So we
check whether this is the case using the isElementVisi-
ble() method. If it’s not visible, the menu’s scroll position is
adjusted using jQuery’s scrollTop() method, which makes
sure it’s in view.
Next, the new option is stored so that it can be referenced

later when the method is called again for a different option.
And finally, the option is focused programmatically to
ensure its value is announced in screen readers.
150 Chapter 3
To provide feedback to sighted users we can use the same

[aria-selected=true] CSS attribute selector like this:
.autocomplete [role=option][aria-selected="true"] {
background-color: #005EA5;
border-color: #005EA5;
color: #ffffff;
}
Tying state and its visual representation together is a good

thing because it ensures that state changes are communi-
cated interoperably. Form should follow function, and doing
so directly keeps them in-sync.
The Basic Filter Function
Having looked at the main interaction flows and the

routines that run off the back of them, we can look more
closely at the filtering mechanism. A good filter is designed
to forgive small typos and letter casing. It’s worth reminding
ourselves again that the data driving the suggestions resides
in the select box <option> elements.
<select>
<option value="">Select</option>
<option value="1">France</option>
<option value="2">Germany</option>
<option value="3">Spain</option>
</select>
As noted above, the getOptions() method is called when

we need to populate the menu with matching options.
Autocomplete.prototype.getOptions = function(value) {
var matches = [];
// Loop through each of the option elements
this.select.find('option').each(function(i, el) {
// if the option has a value and the option’s text node
matches the user-typed value
if($(el).val().trim().length > 0 && $(el).text().
toLowerCase().indexOf(value.toLowerCase()) > -1) {
// push an object representation to the matches array
matches.push({
text: $(el).text(),
value: $(el).val()
});
}
});
return matches;
};
The method takes the user-entered value as a parameter. It

then loops through each of the <option>s and compares the
value to the option’s text content (the bit inside the element).
It does so by using indexOf() which checks to see if the
string contains an occurrence of the specified value. This
means users can type incomplete parts of countries and still
have relevant suggestions presented to them.
152 Chapter 3
The value is trimmed and converted to lowercase, which

means options will still be shown if the user has, for exam-
ple, turned on caps lock on their keyboard. Users shouldn’t
have to fix problems we can fix for them automatically.
Each matched option is added to the matches array, which

will be used by the calling function to populate the menu
accordingly.
Supporting Endonyms and Common Typos
An endonym is a name used by the people from a particu-

lar area of that area (or themselves or their language). For
example, Germany in German is “Deutschland.” We can
follow inclusive design principle 5, “Offer choice,” by letting
users type an endonym.
To do this, we first need store it somewhere. We can put the

endonym inside a data attribute on the <option> element.
<select>

<option value="2" data-alt="Deutschland">Germany</option>

</select>
<select>
With the select box ready, we can change the filter function to
check the alternative name like this:
Autocomplete.prototype.getOptions = function(value) {
var matches = [];
// Loop through each of the option elements
this.select.find('option').each(function(i, el) {
// if the option has a value and the option’s text node
matches the user-typed value or the option’s data-alt
attribute matches the user-typed value
if( $(el).val().trim().length > 0
&& $(el).text().toLowerCase().indexOf(value.
toLowerCase()) > -1
|| $(el).attr('data-alt')
&& $(el).attr('data-alt').toLowerCase().indexOf(value.
toLowerCase()) > -1 ) {
// push an object representation to the matches array
matches.push({
text: $(el).text(),
value: $(el).val()
});
}
});
return matches;
};
The attribute isn’t reserved for endonyms — it can be used to

store common typos too.
154 Chapter 3
How It Might Look
The autocomplete component showing suggestions as the user types.
2. When to Fly
Dates are notoriously hard: different time zones, formats,
delimiters, days in the month, length of a year, daylight
savings, and on and on.11 It’s hard work designing all this
complexity out of an interface.
Often, three select boxes are used: one for day, month, and
year. Admittedly, we’ve just discussed the cons of select
boxes, but it must be said that one of their redeeming quali-
ties is that they help users enter the right information. But
in the case of dates, even this quality doesn’t hold up because
you can select an invalid date, such as 31 February 2017.
11 http://smashed.by/falsehoodstime
A date of birth field made up of three select boxes: day, month and year.
Select boxes are also used to avoid locale and formatting

differences. Some dates start with month, others with day.
Some delimit dates with slashes, others with dashes or dots.
We can’t reliably determine users’ intention based on what
they enter. It’s just one of those things.
A text box populated with “10/09/12” which could be one of several dates
depending on the format.
Many of us assume that using a calendar widget is always

better than letting users type freely into a text box. But this
is not always the case. The GDS Service Manual states:12
12 http://smashed.by/dateselector
156 Chapter 3
The way you should ask users for dates depends on the types of date
“ you’re asking for.
Let’s walk through some of the main types of dates to see

what interface is best for users. Then we can see if any of
those are suited to the context of our problem: choosing a
date to fly on.
DATES FROM DOCUMENTS

Here’s what GDS says about asking for dates found on docu-
ments and other physical items users may need to reference:
If you ask for a date exactly as it’s shown on a passport, credit card
“ or similar item, make the fields match the format of the original.
This will make it easier for users to copy it across accurately.
The expiry date from chapter 2, “Checkout,” falls under this

category. As the expiry date is just four characters with an
optional slash, we gave users a single text box that matches
the expected format. Essentially, users just copy what they
see. Easy.
MEMORABLE DATES
A memorable date is one that you remember easily such as
your date of birth. Typing six digits unassisted into a text
box is much quicker than scrolling, swiping, and clicking
through multiple years, months, and days within a calendar.

Memorable dates are best represented by three text boxes:
one for day, month, and year. Why three? Because it solves
the locale and formatting issues mentioned earlier.
Date of birth field made up of three text boxes: day, month, and year.
<legend>
<span class="field-legend">Date of birth</span>
<span class="field-hint">DD MM YYYY</span>
</legend>
<div class="field-dayWrapper">
<label for="day">Day</label>
<input class="field-dayBox" type="text"
pattern="[0-9]*" name="day" id="day">
</div>
<div class="field-monthWrapper">
<label for="month">Month</label>
<input class="field-monthBox" type="text"
pattern="[0-9]*" name="month" id="month">
</div>
<div class="field-yearWrapper">
<label for="year">Year</label>
<input class="field-yearBox" type="text"
pattern="[0-9]*" name="year" id="year">
</div>
</fieldset>
158 Chapter 3
The three fields are wrapped in a fieldset. The legend

(“Date of birth”) gives each text box context and would be
read out as “Date of birth, day” (or similar) in screen readers.
Note: The pattern attribute is used to trigger the numeric

keyboard — a little enhancement for iOS users. If you’re
wondering about why we haven’t used the number input,
you can refer back to the number input in “A Checkout
Form.”
A DATE PICKER
When choosing a date to fly on, users are neither entering
a memorable date nor one found in a document. They are
searching for a date sometime in the future and usually
within the next few months.
We tend to think of time in structured chunks: days, weeks,

and months, and so on. And we plan our time using calendars
which align with that notion. It’s sensible then, to let users
find and pick a date through a familiar and intuitive calendar
interface, or what’s commonly known as a date picker.
As usual, our first port of call is to see if there’s a date picker

control that browsers provide natively for free. Good news:
there is. The date input (<input type="date">) offers a
special and convenient interface for picking dates while also

enforcing a standard format for the value that’s sent to the
server on submission.
Mobile browser support is really good and includes Sam-

sung’s browser, Firefox, Edge, Chrome, Opera, and Safari.
Desktop support is patchier: Chrome and Edge support it,
but Internet Explorer and Safari don’t (at time of writing).
We’ll look at how to support them later.
A selection of date pickers on different browsers.
As the date picker is provided by the browser, you’ll notice

how it looks a lot like the system date picker that’s used for
setting dates and times on your phone. That’s by design so
that mobile browsers can outsource the problem to native
components. This is good because users will be familiar
with it, which speaks to inclusive design principle 3, “Be con-
sistent.”
160 Chapter 3
You might be concerned that they look different in different

browser vendors. Don’t be. Most users don’t notice the dif-
ference and the rest don’t care. Remember, most people use
the same browser every day. They only see their platform’s
implementation. Unlike us, they’re not agonizing over sub-
tle differences during cross-browser testing.
Nobody cares about your website as much as you do.
“ — Goran Peuc, “Nobody Wants To Use Your Product”13
If you’re not able to conduct your own user research, watch

“Progressive Enhancement 2.0” (at about 29 minutes in).14
Nicholas Zakas shows the audience a slide with a photo
on it. He moves to the next slide which contains the same
photo. He then asks the audience if they noticed any dif-
ferences. Even though the second photo had a border and a
drop shadow, not one person noticed.
Ironically, the audience was made up of designers and

developers — people who are trained to notice these
things. But they didn’t notice them, because like any user,
they were focused on the content, not the finer points of
the visual aesthetic.
13 http://smashed.by/nobodywantsyourproduct
14 http://smashed.by/progenhance
The Basic Markup
This markup has been used many times already in the book.
The only difference is that the input’s type attribute is set to
date.
<div class="field">
<label for="departure">
<span class="field-label">Departure date</span>
</label>
<input type="date" id="departure" name="departure">
</div>
Browsers that support the date input give users a standard

date picker that’s familiar, accessible, and performant by
default. So that’s good. But what happens in other browsers?
The date input will change into a basic text input — this is
known as graceful degradation. While users won’t get the
convenience of a date picker, they’ll still be able to enter
a date. This is an example of HTML’s inherently resilient
nature. When things fails, they don’t break.
Depending on your situation, this level of support may be

fine. Perhaps entering a date is of low priority, or happens
too infrequently to worry about. Or perhaps none of your
users will ever use an unsupported browser. But finding a
date is integral to the flight booking experience.
162 Chapter 3
We ought to provide a better experience for people using an

unsupported browser.
How It Might Look
The date picker consists of a toggle button that reveals the

calendar. From there, users will be able to traverse the calen-
dar and ultimately select a date to populate the text box.
The enhanced date picker interface shown in its expanded state, with a but-
ton to the right of the text field, and the calendar inline below it.
Notes about the Design
Many date pickers are designed as overlays, but they

obscure the rest of the page, and are prone to being cropped
by the viewport when positioned absolutely on top of the
interface. Instead, our calendar will be positioned under-

neath the input and inline, avoiding such issues.
There’s an inset left border which visually connects the

calendar to the field above. And the interactive elements
within the calendar have large tap targets which are easier
to tap and click.15
You might be tempted to try to squeeze additional infor-

mation — such as price and availability — into each of the
cells. This may be possible in very large viewports, but it’s
not practical from a responsive design perspective: there’s
simply not enough room to denote this information in small
viewports. This is why it’s important to design mobile-
first. In any case, the primary user need at this stage of the
journey is to select a date. Trying to squeeze in additional
information is going to result in a slower, busier and over-
whelming experience that slows users down.
Instead, we’ll let users focus on choosing a date unencum-

bered, and later we’ll give users more information when it’s
both useful and practical to do so.
15 http://smashed.by/sizetaptargets
164 Chapter 3
Feature Detection
As we only want to give unsupported browsers the cus-

tom date picker component, the first thing we need to do
is detect when and when not to initialize the component.
Without a provision in place, users might see two opposing
date pickers: the native one, and our own, which would be
confusing and unnecessary.
We can check for support before enhancing the interface

using a little feature detection script.
function dateInputSupported() {
var el = document.createElement('input');
try {
el.type = "date";
} catch(e) {}
return el.type == "date";
}
The function works by trying to create a date input and then

checking to see if its type attribute is correctly reported as a
date input. In browsers that lack support, it will be reported
as a text input instead. We can use this function to deter-
mine whether the DatePicker() component should be
defined or not.
if(!dateInputSupported()) {
var DatePicker = function() {
// code here
};
}
As the DatePicker() is only defined when there’s no support

for the native date input, we can check to see if it’s defined
before initializing it. This is known as a dynamic JavaScript
API, because it changes based on support and is a crucial
aspect of designing progressively enhanced interfaces.16
if(typeof DatePicker !== "undefined") {

new DatePicker();
}
The Enhanced Markup
After the date picker has been initialized, the markup will
have been changed to include the date picker controls:
toggle button, next and previous month buttons, and the
calendar grid.
16 http://smashed.by/featuredetection
166 Chapter 3
<div class="field">
<label for="when">
<span class="field-label">Date</span>
</label>
<div class="datepicker">
<input type="text" id="when" name="when">
<button type="button" aria-expanded="true" aria-
haspopup="true">Choose</button>
<div class="datepicker-wrapper hidden">

</div>
</div>
</div>
Notes
• The type="button" attribute stops the button
from submitting the form. If the type was set to
submit (or omitted altogether) when pressed, it
would incorrectly submit the form.
• The aria-haspopup="true" attribute indicates
that the button reveals a calendar. It acts as a warn-
ing that, when pressed, the focus will be moved to
the calendar. Its value is always set to true.
• The aria-expanded attribute indicates whether
the calendar is currently in an open (expanded) or
closed (collapsed) state by toggling between true
and false values.
Revealing the Calendar
The calendar starts hidden. When the toggle button is

clicked, the calendar is revealed by removing the `hidden`
class on the wrapper.
The date picker in its original state (left), and after revealing the calendar (right).
Immediately after, the focus is set to the Previous Month but-

ton, which is the first focusable element inside the calendar.
DatePicker.prototype. {
// showing
if(this.toggleButton.attr('aria-expanded') == 'true') {
this.hide();
// hiding
} else {
this.show();
this.calendar.find('button:first-child').focus();
}
};
168 Chapter 3
DatePicker.prototype.hide = function() {
this.calendar.addClass('hidden');
this.toggleButton.attr('aria-expanded', 'false');
};
DatePicker.prototype.show = function() {
this.calendar.removeClass('hidden');
this.toggleButton.attr('aria-expanded', 'true');
};
The Calendar HTML
The container has two important attributes: the

role="group" attribute groups the related calendar con-
trols together. When the calendar is revealed, screen reader
users will hear the button’s label in combination with the
group’s label: “date picker, previous month, button.”
<div class="datepicker-calendar" aria-label="date picker"

role="group">
<div class="datepicker-actions">
<button type="button" aria-label="previous month">
<svg focusable="false" version="1.1"
xmlns="http://www.w3.org/2000/svg"
xmlns:xlink="http://www.w3.org/1999/xlink" viewBox="0 0 17
17" width="1em" height="1em">...</svg>
</button>
<div role="status" aria-live="polite">February 2018</div>
<button type="button" aria-label="next month">
<svg focusable="false" version="1.1"
xmlns="http://www.w3.org/2000/svg"
xmlns:xlink="http://www.w3.org/1999/xlink" viewBox="0 0 17
17" width="1em" height="1em">...</svg>
</button>
</div>

</div>
The month’s title and year are placed within a live region (as
first discussed in chapter 2). This means its content will be
announced by screen readers when the calendar is revealed.
This same information will be continually announced as
users move between different months.
Note: As mentioned earlier, the focusable="false"

attribute on the SVG icon fixes the issue that in
Internet Explorer SVG elements are focusable.
Previous and Next Month Buttons
With the calendar now revealed, the user can move between
the Previous Month and Next Month buttons by using the
Tab key. This is because we’ve used <button> elements,
which are naturally focusable and part of the tab sequence.
Buttons are interoperable meaning they can be activated

by clicking, tapping, or pressing Space or Enter with the
keyboard. All we need to do is listen for the click event.
170 Chapter 3
DatePicker.prototype.addEventListeners = function() {
this.calendar.on('click', 'button:first-child',
$.proxy(this, 'onPreviousClick'));
this.calendar.on('click', 'button:last-child',
$.proxy(this, 'onNextClick'));
};
this.showPreviousMonth();
};
this.showNextMonth();
};
The showPreviousMonth() and showNextMonth() meth-

ods (not shown) work out which month to show, and then
update the title and grid HTML accordingly.
The Grid
The days of the month are presented in a grid format of

which the <table> element is perfectly suited, so that’s easy.
But it’s the careful arrangement of attributes that is crucial in
not only enabling interaction but also understanding.
<table role="grid">
<thead>
<tr>
<th aria-label="Sunday">Su</th>
<th aria-label="Monday">Mo</th>
<th aria-label="Tuesday">Tu</th>
<th aria-label="Wednesday">We</th>
<th aria-label="Thursday">Th</th>
<th aria-label="Friday">Fr</th>
<th aria-label="Saturday">Sa</th>
</tr>
</thead>
<tbody>
<tr>
<td tabindex="-1" aria-selected="false"
aria-label="4 February, 2018" role="gridcell"
data-date="Sun Feb 04 2018 00:00:00 GMT+0000 (GMT)">
<span aria-hidden="true">4</span>
</td>
data-date="Mon Feb 05 2018 00:00:00 GMT+0000 (GMT)">
</td>
data-date="Tue Feb 06 2018 00:00:00 GMT+0000 (GMT)">
</td>
data-date="Wed Feb 07 2018 00:00:00 GMT+0000 (GMT)">
</td>
data-date="Thu Feb 08 2018 00:00:00 GMT+0000 (GMT)">
</td>
data-date="Fri Feb 09 2018 00:00:00 GMT+0000 (GMT)">
172 Chapter 3
</td>
<td tabindex="-1" aria-selected="false" aria-
label="10 February, 2018" role="gridcell"
data-date="Sat Feb 10 2018 00:00:00 GMT+0000 (GMT)">
</td>
</tr>
<tr>...</tr>
<tr>...</tr>
<tr>...</tr>
</tbody>
</table>
The role="grid" attribute (and each cell’s role="grid-

cell" attribute) tells screen readers to treat the table as
a grid. Without this, JAWS, for example, won’t let users
operate the calendar with the arrow keys using JavaScript.
This is because the arrow keys are reserved for operating a
standard table in a special way.
The <thead> contains the names of the days. Note that

they’re abbreviated, which should be avoided in most cases.
But the calendar needs to fit on small viewports too. The
unabbreviated heading is placed inside the aria-label
attribute. Support is a little patchy, but it’s a useful enhance-
ment for browser/screen reader combinations that will pick
this up, such as NVDA with Firefox.
Each cell contains a number (the date), which is perfectly

adequate for sighted users as they can see the entire calen-
dar. But screen reader users would only hear “Seventeen,”
which is ambiguous unless they carefully remember the
previously announced month and year. To provide a compa-
rable experience (inclusive design principle 1), we put the full
date inside the aria-label attribute.
The <span> has an aria-hidden="true" attribute, which

stops the number being read out twice by some screen read-
ers, without hiding it from sighted users.
The tabindex, aria-selected and data-date attributes

will be discussed shortly.
Clicking a Day
When the user clicks a day, a number of actions must be

performed. The event handler looks like this:
var d = new Date($(e.currentTarget).attr('data-date'));
this.updateTextBoxDate(d);
this.hide();
this.input.focus();
this.selectDate(d);
this.selectedDate = d;
};
174 Chapter 3
First, the date string (stored inside the cell’s data-date

attribute) is converted into a JavaScript Date object which is
then used to populate the text box. Then the date picker is
hidden and the text box is focused.
The selectDate() method will mark the selected cell by

setting the aria-selected attribute to true, and by setting
the previously selected cell to false.
Finally, the selected date is stored so we can show the

calendar in the correct state if the user decides to pick
another date.
Keyboard Interaction
Like the autocomplete component we designed earlier, the grid

is a composite control made up of many interactive elements
— as many as 31, depending on the month. As discussed earlier,
composite controls should have just one tab stop: having to tab
through 31 days is tiresome and inefficient.
To solve this problem, we’re going to use the concept of

roving tabs.17 The way it works is that only one cell in the grid
is focusable at any one time — the selected cell. The selected
cell will have a tabindex="0" attribute, which means users
can tab to it from the Next Month button.
17 http://smashed.by/tabindex
The rest of the cells have a tabindex="-1" attribute, which

means focus can be set programmatically with JavaScript,
but users won’t be able to get to it with the Tab key.
Once the selected cell is focused, users can traverse the

calendar with the arrow keys. As the user moves between
the cells, the tabindex values will be updated to ensure that
only the selected cell has a tabindex value of 0.
DatePicker.prototype.addEventListeners = function() {
// ...
this.calendar.on('keydown', '[role=gridcell]',
$.proxy(this, 'onCellKeyDown'));
// ...
};
switch(e.keyCode) {
this.onDayDownPressed(e);
break;
case this.keys.up:
this.onDayUpPressed(e);
break;
case this.keys.left:
this.onDayLeftPressed(e);
break;
case this.keys.right:
this.onDayRightPressed(e);
break;
this.onDaySpacePressed(e);
break;
}
};
176 Chapter 3
Note: As described earlier with the autocomplete

component, we could have used the aria-
activedescendant technique for the grid. However,
the benefits of using roving tab indexes is that it’s better
supported and it ensures that the newly focused element is
scrolled into view.
Key Action
Down Focus the same day in the subsequent week. If it’s the
last week, switch to the next month first.
Up Focus the same day in the previous week. If it’s the
first week, switch to the previous month first.
Left Focus the previous day. If it’s the first day of the
month, switch to the previous month first.
Right Focus the next day. If it’s the last day of the month,
switch to the next month first.
Enter or Performs the same actions as clicking the day: popu-
Space late and focus the text box, and hide the menu.
Escape Hide the calendar and focus the toggle button.
Note: While screen reader users can operate the calendar

like this, it’s probably easier and quicker for them to type
a date directly into the text box. But inclusive design is
about not making such assumptions. Instead we let the
individual user decide (inclusive design principle 5).
Doing Our Best
Despite our efforts to support as many users as possible,

there’s a rare situation whereby users still won’t get a date
picker.
In chapter 1 we discussed the importance of progressive

enhancement, because we can’t be sure that JavaScript is
always available. Users experiencing a network or JavaScript
failure while also using a browser that doesn’t support the
native date will just see a text box, without any hint text
regarding the format of the date.
The date picker without a hint, as seen when JavaScript is unavailable.
We can’t use the hint pattern (from chapter 1) because

browsers that support the date input use a different format.
Of course, we should be as forgiving as possible, by letting
users type slashes, periods, or spaces, but typing a two-digit
year first, for example, will still cause an error. A well-writ-
ten error message may have to suffice.
178 Chapter 3
Another option would be to provide a hint via the place-

holder attribute (and remove it when the date picker is
initialized). Despite the many problems with placeholders
(as discussed in “A Registration Form”), this might be the
lesser of two evils.
Design is often a question of priorities. What is a good

experience for most may create a less-than-ideal experience
for some, which is especially the case on the web. Inclu-
sive design is about making decisions that are unlikely to
exclude people.
“ People ignore design that ignores people.

— Frank Chimero
In this rare situation, users are still able to enter a date

which makes this pattern an accessible one. In the end, it’s
about doing our best and we’ve done that here.
Future Support
The web is constantly changing. Browsers and devices are

released at a rapid rate, each with varying features and
capabilities. This is why Jeremy Keith refers to the “web as a
continuum, not a platform.”18
18 http://smashed.by/webcontinuum
We need to think about what level of support makes sense

for our users depending on the feature at hand. Earlier, we
decided to enhance the experience for browsers that don’t
support the native date input, which makes sense today.
As browser support improves, the number of people who

would experience the degraded version will diminish; at
which time we can remove our custom date picker code.
Not only does this give us less to maintain, but users will
get a faster experience as they don’t need to download the
code. Lovely.
3. Choosing Passengers
Next we need to know how many people are travelling. The
age of the passengers affects the price of the ticket, so we’ll
arrange the interface according to these age groups.
Passenger count form with three fields: one for adults, children, and infants.
180 Chapter 3
<div class="field">
<label for="adults">
<span class="field-label">How many people aged 16 years
and over are flying?</span>
</label>
<input type="number" id="adults" name="adults" min="0"
max="9">
</div>
<div class="field">
<label for="children">
<span class="field-label">How many children, aged
between 2 and 15 years old, are flying?</span>
</label>
<input type="number" id="children" name="children"
min="0" max="9">
</div>
<div class="field">
<label for="infants">
<span class="field-label">How many infants, under 2
years old, are flying?</span>
</label>
<input type="number" id="infants" name="infants" min="0"
max="9">
</div>
Each age group is a separate field. As we’re asking users for

an amount of something — passengers — the number input
makes sense. (We discussed when to use the number input
in “A Checkout Form.”)
Number inputs have little spinner buttons (also called step-

pers), which let users increase or decrease the input’s value
by a constant amount. Luke Wroblewski’s usability testing
shows that users prefer them to drop down menus:
When testing mobile flight booking forms, we found people
“ preferred steppers for selecting the number of passengers. No

dropdown menu required, especially since there’s a maximum of 8
travelers allowed and the vast majority select 1–2 travelers.
The only downside is that the browser-provided spinners

are tiny, which make them difficult to use. And some brows-
ers don’t show them at all. We can solve this problem by
creating our own custom stepper component.
A STEPPER COMPONENT
To supply all browsers with bigger, more ergonomic but-
tons, we can create a custom stepper component using
JavaScript. On mobile, they’ll save users from triggering the
on-screen keyboard, which reduces the time and effort to
complete the task.
182 Chapter 3
How It Might Look
Enhanced passenger form with stepper buttons, making it simple to add or

remove passengers.
Hiding the Native Spinners
But first, we need to turn off the native, browser-provided

spinners. In WebKit browsers we can hide them like this:
input::-webkit-outer-spin-button,
input::-webkit-inner-spin-button {
-webkit-appearance: none;
appearance: none;
margin: 0;
}
The Enhanced Markup
When the Stepper() component initializes, the markup

will be changed to this:
<div class="field">
<label for="adults" id="adults-label">How many people
aged 16 and over are flying?</label>
<div class="stepper">
<button type="button" aria-label="Add" aria-
describedby="adults-label">−</button>
<input type="number" id="adults" name="adults"
value="1">
<button type="button" aria-label="Remove" aria-
describedby="adults-label">+</button>
<div class="visually-hidden" role="status" aria-
live="polite">1</div>
</div>
</div>
Notes
• The buttons and number input are wrapped in a
<div> so they can be styled as a group underneath
the label.
• The button’s aria-label attribute ensures that
screen readers announce “Remove” instead of “mi-
nus symbol.” Same goes for “Add” instead of “plus
symbol.”
184 Chapter 3
• The button’s aria-describedby attribute referenc-

es the label’s id, which means it combines with the
label text to give screen reader users extra context.
As there are three fields on the page, this stops us-
ers thinking “Remove — remove what, exactly?”
• Each button has a type="button" attribute to stop
the form submitting when clicked.
• Clicking the Add or Remove buttons updates the
live region so screen reader users will hear the
change without having to move away from the
button (see note).
Note: When the Add (or Remove) button is clicked, the

input’s value is updated, but screen readers don’t announce
this change. At first, I put the live region attributes on the
input. This didn’t work in some screen readers, but worse
was that it changed the input’s semantics into a status box.
A Note on Using Iconography
You’ll notice that we’re using icons for the buttons. Icons
are often the source of heated debates amongst designers,
mostly because they have their pros and cons.
The pros are that they save space, and internationally

recognized icons overcome language barriers, which is why
they’re often used in airports.
The cons are that icons can be misunderstood, and they are
a poor replacement for just using text. In “The best icon is a
text label,” Thomas Byttebier goes as far to say:19
What good has a beautiful interface if it’s unclear? Hence it’s simple:
“ only use an icon if its message is a 100% clear to everyone. Never

give in.
In the case of the stepper buttons, plus and minus icons

keep the options equally weighted and are widely under-
stood. Moreover, users can type a number if they want,
ignoring the buttons altogether.
19 http://smashed.by/texticons
186 Chapter 3
4. Choosing a Flight
Now all the relevant information has been collected, we can
give users a list of flights from which to choose one.
Flight chooser form made up of radio buttons containing departure time,

arrival time, and price. The cheapest price is marked with a “Best price” label,
and there are pagination controls at the bottom to reveal other days’ flights.
The system shows flights that match the date the user spec-
ified earlier. Additionally, the interface lets users move back
and forth between days. The group’s label is set as normal via
the <legend> and reads “Available flights on 18 August 2018.”
The flights are represented as radio buttons — the user can

select only one. Each label contains the departure time,
arrival time, and ticket price: all useful information. One
advantage of using radio buttons is that you can add any

information inside the label and style it as you like, some-
thing you couldn’t do if you were using a select box.
<label for="flight1">
<input type="radio" name="flight" value="1"
id="flight">
<span>Departing at 18:20pm</span>
<span>Arriving at 20:30pm</span>
<span>£99</span>
</label>
</div>
GROUP VALIDATION ERRORS

In chapter 1 we looked at how to design an inclusive form
validation experience. But because the registration form only
consisted of two simple text fields, we never looked at how to
handle errors for a field consisting of multiple form controls.
A radio button group is made up of multiple controls. Take

a look at the markup below. The fieldset contains the group
of controls, and the legend is the group’s label. We can
effectively use the same error pattern by injecting the error
<span> inside the legend. Not only will sighted users see
the error, but screen reader users will hear the error too.
188 Chapter 3
<fieldset aria-invalid="true">
<legend>
<span class="field-legend">
Available flights on 18 August 2018
</span>
<span class="field-error">
xlink:href="#warning-icon"></use></svg>
Choose a flight.
</span>
</legend>
<input type="radio" name="flight" value="1" id="flight">
...
</label>
</div>
<input type="radio" name="flight" value="2" id="flight2">
...
</label>
</div>
<input type="radio" name="flight" value="3" id="flight3">
...
</label>
</div>
</fieldset>
The aria-invalid="true" attribute is placed on the

fieldset. Putting it directly on the radio button would
be incorrect here, because it’s not the individual input
that’s invalid — it’s the group. The error <span> is exactly
the same as the one used for standard text fields, which
ensures that errors look and behave the same across all
types of form fields, which speaks to inclusive design princi-
ple 3, “Be consistent.”
Flight radio buttons with error message.
The error summary needs to contain a link to the first radio

button within the group. That is, the link’s href attribute
needs to match the first radio button’s id attribute. This is
why the first radio button in the group has matching id and
name attributes: “flight.”
190 Chapter 3
<div class="errorSummary" role="group" tabindex="-1" aria-

labelledby="errorSummary-heading">
<h2 id="errorSummary-heading">There’s a problem</h2>
<ul>
<li><a href="#flight">Choose a flight</a></li>
</ul>
</div>
5. Choosing A Seat
Choosing a seat isn’t the most complicated part of the jour-
ney, yet the combination of perceived affordance, layout,
and interaction design can make or break this part of the
journey if we’re not careful.
LAYOUT
Up to now, any field that uses radio buttons has them
stacked beneath one another, which is good for most situ-
ations. For the seat chooser, however, this makes the page
especially long, and — more importantly — harder to scan,
as there’s a lack of structure.
We can provide that structure by laying out the seats in

rows, just like they are on a plane. This will help users map
their location, which is useful because users might be look-
ing for aisle or window seats, for example.
Left: seat checkboxes stacked beneath each other making the page
long. Right: seat checkboxes laid out in rows making the page shorter
and seats easier to find.
To demarcate window seats and aisle seats for screen reader

users, we can put hidden text inside the seat’s label.
<label for="S1A">
<input type="checkbox" name="seat" value="1A" id="S1A">
<span class="plane-seatNumber">1A <span
class="vh">Window</span></span>
</label>
NESTED FIELDSETS
The radio buttons are placed inside an extra fieldset (and
legend) to indicate which class the seat belongs to: first
class, or economy. Visually this is fine, but screen readers
don’t always behave as expected. Sometimes, they announce
both legends when the first radio button is focused. Some-
times they don’t announce the outer legend at all. You can
192 Chapter 3
read Léonie Watson’s article “Using the fieldset and legend

elements” for more information about this.20
Where possible, you should avoid nested fieldsets, not only

for screen reader users, but also because their existence
often signifies extra complexity that can be designed better
with a little more thought. In our case, we’re showing both
first class and economy class seats because users were never
asked to specify which class they wanted earlier in the
journey.
Instead, we can ask users to specify their preference before-

hand. At the same time, we can mark “Economy class” as
checked by default. Marking the most common choice
expedites the process.
Class chooser with two radio buttons: economy, and first class.
20 http://smashed.by/fieldsetelements
CHECKBOXES ARE NEVER ROUND

In “Checkboxes Are Never Round,” Daniel De Laney says:21
Interactive things have perceived affordances; the way they look
“ tells us what they do and how to use them. That’s why checkboxes
are square and radio buttons are round. Their appearance isn’t just
for show—it signals what to expect from them. Making a checkbox
round is like labeling the Push side of a door Pull.
A radio button tells you that just one can be selected; check-
boxes tell you that more than one can. So if one person is
travelling, use radio buttons; otherwise, use checkboxes.
OOPS, WE BROKE THE RULES

Miller’s law would have you believe that we shouldn’t
present users with more than seven radio buttons at a time.
If you have more than seven, traditional advice would be to
use a select box.22
Laws are useful: they work as constraints that drive us to

good, creative solutions; they allow us to think less, free up
our time to solve other problems, and avoid mistakes others
have made in the past. But UX Myth 23 states that:23
21 http://smashed.by/checkboxes
22 http://smashed.by/millerslaw
23 http://smashed.by/sevenchoices
194 Chapter 3
Miller’s original theory argues that people can keep no more
“ than 7 (plus or minus 2) items in their short-term memory. On

a webpage, however, the information is visually present, people
don’t have to memorize anything and therefore can easily manage
broader choices.
In our case, a Boeing 747 has over 400 seats. Call me a rebel,
but I’m struggling to see a better way of presenting fewer
seats. Choosing a seat is quite a unique interaction and ben-
efits from presenting this many choices in plain site.
Also, using the one thing per page pattern (introduced in

chapter 2) gives us maximal screen space to design some-
thing better. The screen, while long, works well because it’s
dedicated to just one thing: choosing a seat.
UNAVAILABLE SEATS
Unavailable seats are marked by disabling the checkbox (or
radio button). Browsers will gray them out so that sighted
users know they aren’t selectable. Similarly, screen read-
ers won’t announce them, and keyboard users can’t focus
them. This is one of the few use cases where disabling
elements is appropriate.
<input type="checkbox" name="seat" value="1A" disabled>

LAYOUT ENHANCEMENTS
Laying out seats in rows can cause seats to wrap in small
viewports, which destroys the idea of laying them out as
modeled in real life. We could style the seats so they don’t
wrap, but this would cause horizontal scrolling. Neither of
these problems are deal breakers, but if we could reduce the
chance of this happening, we should.
One approach involves hiding the checkboxes and styling

the label to look clickable (which it is). Hiding checkboxes
with CSS alone is dangerous because pressing Tab moves
focus to the checkbox, not the label. On its own this breaks
the interface for sighted keyboard users because as the user
focuses each checkbox there’s no feedback.
To fix this problem, we can give the (still) visible <span

class="plane-seatNumber"> the appearance of focus
using the adjacent sibling selector like this:
.enhanced [type="checkbox"]:focus + .plane-seatNumber {

border: 3px solid #ffbf47;
}
Note that .enhanced is part of the selector. This is because

these styles should only be applied when JavaScript is
available. This is done by adding a class of enhanced to the
document element in the <head> of the document like this:
196 Chapter 3
<script>
document.documentElement.className = 'enhanced';
</script>
LIMITING SELECTION
If the user specified two travellers, we need a way to allow
users to select only two seats. There’s no way natively to limit
the amount of checkboxes the user can check. If a user selects
more than their quota, they’ll get an error message. Without
user research, it’s hard to know whether this is a problem. But
if it is, we can enhance the experience with JavaScript.
One way to do this is to disable the remaining seats as soon as

the limit is reached. But this assumes users will pick the right
seat the first time. When the user tries to click another seat,
the interface won’t respond because that seat will be disabled.
Savvy users may realize they have to deselect the currently

selected seat first, but why should they have to? And what
about less savvy users? As designers, we should do the hard
work to make it simple for users.
If a user surpasses their quota, the currently selected seat

should be unchecked automatically for them. Here’s a little
script to do it.
function SeatLimiter(max) {
this.max = max;
this.checkboxes = $('.plane-seat input');
this.checkboxes.on('click', $.proxy(this,
'onCheckboxClick'));
}
SeatLimiter.prototype. {
var selected = this.checkboxes.filter(':checked');
if(e.target.checked && selected.length > this.max) {
selected.not(e.target)[0].checked = false;
}
};
When a checkbox is clicked, the onCheckboxClick method

is called. The method first checks to see if the checkbox has
been checked or not. If it has, it checks to see if the quota
has been surpassed. If both conditions are true, the previ-
ously selected checkbox is unchecked.
Summary
In this chapter, we continued to use the one thing per page
pattern which allowed us to make use of the total screen
estate. We looked at ways of reducing friction, not only
through interface design, but also by looking at the journey
as a whole.
198 Chapter 3
As much as we tried to use native form controls in their

standard format, it became apparent that custom compo-
nents were necessary to give users the best experience. In
the end we designed four custom components:
• An autocomplete control to let users search through a

long list of destinations quickly and accurately in a way
that matches what they know.
• A date picker component to let users find a date in the
future without having to worry about formatting issues.
• A stepper component to let users make small adjust-
ments to an amount of passengers effortlessly.
• A seat chooser to make seat selection simple, even on
small viewports.
THINGS TO AVOID
• Using radio buttons that look like checkboxes (or vice

versa).
• Using select boxes when better alternatives exist.
• Letting users do the hard work when the interface can
be designed to do the hard work for them.
• Nested fieldsets.
Demos
• Autocomplete:
http://smashed.by/autocompletedemo
• Memorable date:
http://smashed.by/memorabledatedemo
• Date picker:
http://smashed.by/datepickerdemo
• Stepper:
http://smashed.by/stepperdemo
• Seat chooser (nested):
http://smashed.by/seatchoosernesteddemo
• Seat chooser:
http://smashed.by/seatchooserdemo
200 Chapter 4
A Login Form
“As a user, I want to log in to [your service] so that I can [do stuff]”
— Nobody, ever!
N
obody wants to log in to your site. They’re forced
to as a security measure. Without it, everyone has
access to everyone else’s stuff. Bad.
Given how long login forms have been around for and how
basic they are in appearance, you’d be surprised at how
often they contain the same usability mistakes that stop
users doing something they don’t even want to do in the
first place. Add social login into the mix and things get
even harder.
In this chapter, we’ll design a login form, and as we bump

into each of the problems, we’ll look at ways to remedy
them. By process of elimination, users should be left with a
straightforward and relatively pleasant login experience.
A Login Form 201
A Standard Login Form
A basic login form with username and password fields.
Username Label and Hint Text

Our login form, like many on the web, has an ambiguous
label of “Username,” even though it expects users to enter
their email address. Ultimately, the login form should mir-
ror the site’s registration form. In our case, this means the
label should be “Email address.”
202 Chapter 4
A field labeled “Username” (left) and “Email address” (right).
Legacy systems sometimes let users enter an email

address or a username. In this case, the same rules apply
— the label should be “Username or email address” — don’t
make users guess.
A field labeled “username” (left) and “username or email address” (right).
Some niche sites, such as airlines, ask users to enter their

booking reference number. In this case, use the hint pattern
to tell users where they can find it.
The booking reference field without hint (left) and with hint (right).
A Login Form 203
Auto-Capitalization, Autocorrect
and Spell-Checking
Some Android and iOS browsers try to help users by auto-cap-
italizing words in text boxes (<input type="text">). For
example, if I type “adam,” it will be changed to “Adam,” which
can be helpful depending on the circumstance.
Prior to iOS 5, this behavior also applied to the email input.

In the case of the username or email address, we certainly
don’t want users to exert energy fixing mistakes they didn’t
even make. So we can disable this behavior like this:
<input autocapitalize="none">
Similarly, iOS will autocorrect words in a text input that it

thinks are mistakes. Continuing with the example above:
a username may contain a random string of characters
that may look like a mistake but isn’t. You can disable
this behavior like this:
<input autocorrect="off">
By the same token, some browsers will mark misspelled

words with an underline. Again, you can disable this:
204 Chapter 4
<input spellcheck="false">
Here’s the final HTML for our email address field:
<div class="field ">

<label for="email">
<span class="field-label">Email address</span>
</label>
<input type="email" id="email" name="email" value=""
autocapitalize="none" autocorrect="off" spellcheck="false">
</div>
Password Field Design

People often use the same password for different sites and
applications. But password rules differ from site to site.
Some sites ask for a capital letter, others ask for numbers
and symbols; other sites ask for a combination of all three.
Many users will tweak their password to match the rules

of the site in question. For example, if their password is
“password,” and the site requires a capital letter, they’ll just
capitalize the first letter to “Password.” Obviously, this is not
recommended, but shows that users usually take the path of
least resistance.
A Login Form 205
Referring back to the registration form in chapter 1 again,

we gave users a hint that explained the password rules. But
like many sites, our login form fails to provide the same
clarity. Why should users have to guess or, worse, reset
their password?
Password field without hint (left) and with hint (right).
This sort of ambiguity is often in the name of security

because providing a hint would make a hacker’s job easier.
But first, hackers don’t hack this way and even if they did,
what’s to stop the hacker checking the rules on the registra-
tion page? Nothing.
Let’s reuse the patterns in “A Registration Form”:
1. Give users a hint text. Users will have a greater chance

of success without having to wait for a useful error
message.
2. Let users reveal their password using the password

reveal pattern (see page 39).
206 Chapter 4
Password field with password reveal component added.
Auto-Tabbing
Some login forms, such as those found on bank sites, ask
users for certain characters of their password. Or they may
ask for certain digits from a security pin. In either case, users
are normally given three separate text boxes or select boxes.
Santander bank password field with separate three text boxes for each
requested character.
The first problem with this approach is that sites will auto-
tab between the fields. That is, focus is moved to the next
text box automatically as the user enters a predetermined
number of characters. But as the BBC’s UX guidance says:1
1 http://smashed.by/managingfocus
A Login Form 207
It can be disorienting and hinder users from verifying information
“ or correcting mistakes if the focus automatically changes when the

user is not expecting it.
Léonie Watson, accessibility expert and screen reader user,

finds them problematic too:2
I strongly dislike having auto-tab functionality imposed on me. It
“ is unexpected, and based on a flawed assumption that it is helpful.

[…] it takes me more time and effort to correct mistakes caused by
auto-tab, than it does to move focus for myself.
This point of view shouldn’t be surprising given the technique

is founded on assumptions that not only break convention, but
also take control away from the user (inclusive design principle 4).
In this case, there’s just no good reason for it. And splitting
up a text box into three is unnecessary. A single, clearly
labelled text box lets users type three characters freely.
Password field with three separate text boxes (left) and a single text box (right).
2 http://smashed.by/autotabbing
208 Chapter 4
Submit Button Text: Log In versus Sign In

Having ironed out problems with the username and pass-
word fields, our login form is almost identical to the regis-
tration form. It contains the same fields in the same order
with the same microcopy. The only difference is the button’s
label. Instead of “Register” it’s “Sign in.”
“Sign in” is perhaps more human than “Log in.” When you
visit a spa or office building, signing in grants you entry.
And you sign out as you leave. It’s usually sensible to use the
same language for digital experiences too.
It can, however, depend on the industry. Banks, for example,

tend to use “Log in.” The notion of logging came along with
computers in the 80s — the operations that users do are
logged for security reasons.
Button labeled “Log in” (left) and “Sign in” (right).
We should design interfaces that speak in a language

familiar to the user. Whichever you choose, be consistent
(inclusive design principle 3). Make URLs, links, headings, and
buttons match. And if users click “Log in” to log in, then
they should click “Log out” to log out.
A Login Form 209
The ‘Username or Password Doesn’t

Match’ Problem
Sometimes, we deliberately make things difficult to use.
A door by its very nature isn’t easy to use — it would be
far easier if there was no door at all. Login forms need to be
somewhat difficult to use, otherwise they wouldn’t be secure.
Many login forms make the same mistake of showing users

an error message that says “The username and password
don’t match.” But as Jared Spool comically explains in “Is
Design Metrically Opposed?”:3
We know which one doesn’t match, we’re just not going to tell you,
“ because our security people think that if we told you that it was the
password, they would know they had a legal username and they
would try every possible password in history.
As noted earlier, hackers don’t actually do this. But even if

they did, it’s easy to check username availability by trying to
register an account with that username.
The problem for users, is that they’re left to reset their

password, which is long-winded and may cause abandon-
ment. Even where a lack of usability or understandability is
deliberate, there still needs to be a degree of understandabil-
ity and usability.
3 http://smashed.by/onelineofcode
210 Chapter 4
In this case, we don’t have to tell users what their password

is, but we can tell them that it’s the password that doesn’t
correspond to the username (which they have right).
Don’t: password error message “The username or password don’t match.”
Do: password error message “The password doesn’t correspond to your username.”
The Form in Context

We’ve ironed out many of the issues surrounding standard
login forms, but we’ve done so while zoomed in on the form
itself. We also need to consider the form in the context of the
page and the overall experience. This includes looking at vari-
ous journeys through the login form. Let’s start with layout.
A Login Form 211
LAYOUT
When trying to perform an action anonymously that
requires being logged in, users will be sent to the login page.
Many sites design the login page with a unique, minimalist

layout. For example, when users try to add a product to their
basket on Tesco’s website, they’re taken to a login page with
a very different layout.
Left: Tesco product list page. Right: Tesco login page with a different layout.
Giving users a different layout is disorienting, especially for

screen reader users and cognitively impaired users, as they
have to familiarize themselves with a new structure.
Where possible, the login form should inherit the layout of

the rest of the site.
212 Chapter 4
ONE FORM PER PAGE

Some sites put both registration and login forms on one page,
either next to each other on desktop, or below each other on
mobile. This is problematic for a number of reasons.
Page containing login and register forms.
• Putting similar forms next to each other makes it hard

to decide between them, especially for cognitively
impaired users.
• Arriving on a page containing two forms, with a
heading of “Log in or register,” is confusing when you
consider that many users would have clicked a link
labelled “Log in.”
• On mobile, one of the forms will be off-screen and
effectively deprioritized.
A Login Form 213
• Screen reader and keyboard users are going to have

to wade through more of the interface to get to the
relevant form.
Instead, put each form on a separate page, and give users a

link to each form at the top.
Left: the login page with a link to the register page. Right: the register page
with a link to the login page.
FORGOTTEN PASSWORD LINK PLACEMENT

Human beings are forgetful. Password managers mitigate
this problem by storing all your passwords in one place —
you just have to remember a single, master password. That’s
great, but password managers aren’t infallible. If you don’t
remember to save your credentials into it, you’re in the same
position as everyone else. Moreover, not everyone uses one,
nor should they have to.
Most sites give users a way to reset their password if they

forget it. The feature itself isn’t especially problematic. It’s
214 Chapter 4
the placement of the link within the login form that can
cause usability issues. If the link is just above the password
field, when users tab from the email field, it’s the link that
will receive focus, not the password field. Some users will
tab and start typing, not realizing what’s happened.
Worse still is when the link is placed before the submit but-
ton. When keyboard and screen reader users tab from the
password field and press Enter, they’ll expect the form to
submit. But instead, they’ll be taken to reset their password.
When they realize what’s happened, they’ll need to go back,
reenter their credentials, and be careful not to make the
same mistake again.
Left: forgot password link between last field and submit button. Right: forgot
password link before the form.
A Login Form 215
When the feature is considered in isolation, having the

reset password link in close proximity to the password field
makes sense. But the primary need is to log in, and the link
shouldn’t disturb the experience of logging in.
The submit button should be the last interactive element

in the form because that’s what users expect. Solving this
problem is simple: place the forgotten password link before
the form, which makes it easy to discover, especially for
screen reader and keyboard users.
Social Login
Sites have recently started to offer users the ability to log in
with social networks, such as Facebook, Twitter, and Google.
This saves users having to type credentials they may not
remember.
Medium lets users login in with Facebook.4 This is a boon

for Medium users because they’ll then have the option to
post articles to Facebook automatically.
Social login is not without its problems though.
4 https://medium.com
216 Chapter 4
PRIVACY
Users are worried about their privacy. They don’t know
what you’ll do automatically, and they want to feel as
though their information is safe and any actions they per-
form are intended.
Medium does this well: on the login page it says, “We won’t
post without asking,” which puts users’ minds at ease.
Medium’s social login buttons explaining “We won’t post without asking.”
SEAMLESS INTERCHANGE
Some users won’t remember how they originally cre-
ated an account; therefore, they won’t know which login
method to choose.
At Kidly, we handled this by showing users an error message.

For example, if users had signed up with standard login, then
tried to sign in with social login, we’d show an error message
saying so. But this puts the burden on the user.
A Login Form 217
Again, Medium lets users log in interchangeably without ever

knowing what happened. For example, if I log in to Medium
with my email but have registered previously with Facebook,
Medium logs me in automatically and merges my accounts.
Users can visit the settings page to see what accounts are
hooked up, which keeps users informed and in control.
Medium’s settings page lets users connect and disconnect different social
media accounts easily.
CHOICE VERSUS CHOICE PARALYSIS

Standard thinking is that choice is good. It offers freedom,
autonomy, and personal responsibility. Heck, even one of the
design principles is “Offer choice.”
218 Chapter 4
But more choice is not necessarily better when it comes to

features or products. Barry Schwartz, author of The Paradox
of Choice, presents a case study in which researchers set up
two displays of jams at a gourmet food store. Customers
could try samples, and were then given a coupon for a dollar
off if they bought a jar. One display had 24 jams, the other
had just 6. Around 30% of people exposed to the smaller
selection bought a jam, but only 3% of those exposed to the
larger selection did.
There’s also some interesting data on companies that offer

pension plans to their employees. One of Barry’s colleagues
was granted access to the records of Vanguard, a mutual
fund company, and found that for every 10 mutual funds
the employer offered, the rate of participation went down
2%. Consider that employees knew that by not participating,
they were passing up as much as $5,000 a year.
This phenomenon is actually called Hick’s law (named after

psychologist William Edmund Hick), which states that the
time taken to make a decision increases as the number of
choices expand.
The point, of course, is that we need to be wary of giving

users multiple ways to log in. It might seem useful, but it
may also be a burden on them. We have to balance the value
in doing so.
A Login Form 219
Summary
In this chapter we started by quashing traditional advice
that omitting hint text and explicit error messages improve
security on login forms. We then looked at some of the sub-
tle usability issues that can be introduced with social login.
Finally, we looked at ways of improving the experience for
keyboard and mobile users, which meant avoiding auto-tab-
bing, autocorrecting and auto-capitalizing input.
THINGS TO AVOID
• Using ambiguous microcopy and error messages in

the name of security.
• Putting the login form next to the registration form.
• Auto-tabbing between multiple fields.
• Using multiple text boxes for one field.
• Putting the forgot password link inside the form.
• Enabling autocorrect, auto-capitalize and spell
check on fields that may not expect real words:
username, for example.
Demo
• Log in form: http://smashed.by/loginformdemo
220 Chapter 5
An Inbox
M
y sister loves to-do lists. In fact, she loves them
so much, that one of her favourite things is mak-
ing new lists out of old ones. The world is full
of lists. There’s even a list of great people.1 On the web there
are several types of lists, and there are some design patterns
that have emerged over the years that help to manage them.
In this chapter, we’ll look at an inbox; that is, a list of emails

sent from other people. In many respects, an inbox is a
list of tasks organized around emails. Besides reading and
replying to them, the aim is to achieve a zen-like state of
Inbox Zero.2 To let users get there quickly, we will design
the interface so they can delete, archive, and mark emails
as spam. But not just one at a time — in bulk. My sister
loves pen and paper, but if we get this right, I hope she’ll be
converted to digital.
List Types
First, we’re going to look at how best to mark up a list of
emails. Discussing lists may seem out of place in a book
1 http://smashed.by/thegreat
2 http://smashed.by/inboxzero
An Inbox 221
about forms, but forms rarely form part of an interface on

their own. Ignoring their surroundings can result in dis-
agreeable experiences.
The meaning (or semantics) behind elements should influ-

ence their appearance. In other words, form should follow
function. There are four elements we can use to construct
lists, each with different semantics: description lists, tables,
ordered lists, and unordered lists. Let’s discuss the pros and
cons of each now.
DESCRIPTION LISTS
A description list (<dl>) — formerly called a definition list
— is for grouping a list of terms and corresponding defini-
tions; for example, product details such as size, price, and
material. As a list of emails isn’t a collection of terms and
definitions, this type of list isn’t appropriate.
<dl>
<dt>Size:</dt>
<dd>250cm × 135cm × 90cm</dd>
<dt>Price:</dt>
<dd>£429.95</dd>
<dt>Material:</dt>
<dd>Reclaimed teak</dd>
</dl>
222 Chapter 5
TABLES
A table (<table>) is an arrangement of data, laid out in rows
and columns. Like a spreadsheet, tables are well-suited for
data that needs to be compared, sorted, and totalled.
Unfortunately, tables are difficult to style on small viewports,

because there’s no room to show more than two or three col-
umns at a time. Even then, it could be a squeeze depending on
the data inside the cells, creating layout issues. For example,
content could wrap profusely, or it could cause users to scroll
horizontally to reveal the hidden content.
Making tables responsive isn’t the most straightforward

thing to do because they are inherently tied to the way they
look. Put another way, to make a table not look like a table
is not only very difficult, but it would be deceptive, counter-
productive, and inaccessible.
Gmail uses tables and puts recipient, subject, and date sent
into columns. Interestingly, though, there are no table head-
ings, which is the first clue that tables have been used for
layout purposes rather than their semantic qualities, which
causes various access issues. Jeremy Keith talks about this
in his book Resilient Web Design:3
3 http://smashed.by/tablelayout
An Inbox 223
Using TABLEs for layout is materially dishonest. The TABLE element
“ is intended for marking up the structure of tabular data. The end

result [...] is a façade. At first glance everything looks fine, but it
won’t stand up to scrutiny. As soon as such a website is stress-tested
by actual usage across a range of browsers, the façade crumbles.
<table class="inbox">
<a href="/email/1">
<tr>
<td>John Oates</td>
<td>Your Amazon.co.uk order #123 is out for
delivery</td>
<td>10 August</td>
</tr>
</a>
</table>
See the markup above as an example. The <tr> is wrapped

in an <a> to let users click an email to read it. The problem
is that browsers ignore the link. It’s simply not allowed and
screen reader users will struggle to interpret it.
Gmail makes the row clickable by using JavaScript to listen

to click events. But not only is this unclear for screen reader
users, not everyone has JavaScript and, quite frankly, it’s
unnecessary.
224 Chapter 5
ORDERED AND UNORDERED LISTS

The generic list is useful because it itemizes and groups
content into related chunks accessibly. But they can be used
for more than just bullet points. They come in two flavors:
ordered (<ol>) and unordered (<ul>) lists. And they’re far
less opinionated than tables.
The difference between the two types lies in their name.

If the order of the items matters, use an ordered list. For
example, a recipe’s instructions require users to follow them
in order — not doing so may produce inedible food. On the
other hand, an inbox doesn’t have to be read or actioned in
a predefined order. It sounds simple when put like that, but
we tend to overthink these things.
Let’s lay out the inbox using a <ul>.
<ul class="inbox">
<li>
<a href="/emails/1/">
<div class="inbox-recipient">John Oates</div>
<div class="inbox-subject">Your Amazon.co.uk order
#123 is out for delivery</div>
<div class="inbox-date">10 August</div>
</a>
</li>
</ul>
An Inbox 225
Note: Unfortunately, most screen readers don’t differentiate

between unordered and ordered lists in any meaningful
way. But this shouldn’t stop us from using the most
appropriate element. As support improves, the benefits will
be ready and waiting.
Unlike tables, unordered lists are stylistically malleable and,

therefore, responsive. That’s because they can be laid out in
different ways without having to affect their structure in a
way that makes them less accessible non-visually.
With the markup above, on large viewports we can lay the

emails out in columns, and on small viewports we can avoid
layout issues by stacking them vertically. Moreover, the
entire list item can be made clickable without resorting to
JavaScript hacks. Wrapping a link around the contents is
perfectly valid, which is less work and more robust.
In the case of an inbox, list items are more suited anyway:

not only are column headings redundant, but there’s no
need to compare or total items in the list.
Marking Email for Action

To let users mark emails for action, we need to give each
row a checkbox.
226 Chapter 5
<ul class="inbox">
<li>
<input type="checkbox" name="email">
<div class="inbox-subject">Your Amazon.co.uk order
#123 is out for delivery</div>
<div class="inbox-date">10 Aug</div>
</a>
</li>
...
</ul>
You’ll notice each checkbox is missing a label. The problem

is that the contents of the link should also be the contents
of the label. In other words, two opposing interactions need
to occupy the same space. Remember, clicking the label
should mark the checkbox, whereas clicking the link should
navigate the user to the email.
In this case, you could argue that a visible label is redundant.

After all, the label would duplicate the link’s content which
would make the experience confusing for sighted users.
USING MODES
Trying to meet two user needs (viewing and managing) in
a single interface is partially responsible for the problem in
the first place. One way to avoid the issue would be to split
these needs apart using the concept of modes.
An Inbox 227
This just means letting users switch between managing

email and reading it.
Top: inbox in read mode where each row is a link to read the email. Bottom: in-
box in manage mode where each row is a label that toggles the checkbox state.
Clicking “Manage” puts users into manage mode. When in

manage mode, the link’s label changes to “Exit” (or similar),
which, when clicked, takes the user back to read mode.
When in read mode, there are no checkboxes or any other

form paraphernalia. The row is a link which takes users to
read the email. When in manage mode, the row turns into a
<label>. When clicked, it marks (or unmarks) the checkbox
just like normal.
228 Chapter 5
Modes are best suited when one mode is used more fre-
quently than the other. But when both are used frequently,
like an inbox, having to switch back and forth all the time
may be undesirable.
Note: We should try to avoid modes, but if they’re

necessary, the interface must make it obvious which mode
is invoked.
VISUALLY HIDE THE LABEL

Instead of using modes, we can add a visually hidden
label. There are two ways to do this. The first is to use the
aria-labelledby attribute (shown below), which uses
existing content to label the checkbox. The downside is that
it means adding id attributes. In any case, ARIA shouldn’t be
used unless there’s no better alternative — something first
noted in chapter 1, “A Registration Form.”
<li>
<input type="checkbox" name="email" aria-
labelledby="inbox_label1">
<a href="/emails/1/" id="inbox_label1">
<div class="inbox-subject">Your Amazon.co.uk order #123
is out for delivery</div>
</a>
</li>
An Inbox 229
Alternatively, a standard <label> has better support and

adheres to ARIA’s first rule (not to use it if a native option is
available). But the downside with this approach is that the
content has to be duplicated, which would create redun-
dancy for sighted users with missing CSS.
<li>
<input type="checkbox" name="email" id="email1">
<label for="email1" class="visually-hidden">From John
Oates, subject ‘Your Amazon.co.uk order #123 is out for
delivery’ (10 August 2017)</label>
<div class="inbox-subject">Your Amazon.co.uk order #123
is out for delivery</div>
</a>
</li>
Note: The CSS for the visually hidden class is set out in “A
Checkout Form.”
While duplication isn’t a big performance issue, if we’re not

careful, bloated HTML can eventually diminish the expe-
rience by causing some operations to take longer — screen
reader software can be unresponsive, for example.
On the other hand, duplication in this case can be advanta-

geous. As the label content is just for screen reader users, we
230 Chapter 5
can create a specific message just for them. For example, the
label has the word “subject” prefixed, which is useful in this
context. This follows inclusive design principle 1, “Provide a
comparable experience,” which is not about giving users the
same experience, but one of comparable value and utility.
HIGHLIGHTING MARKED EMAILS

The deal with human–computer interaction is that when
the human does something, the computer should respond.
In this case, clicking a checkbox makes a little tick appear
(and disappear) accordingly. As with every other checkbox
in any other form, this is probably enough feedback.
Left: checkbox checked. Right: checkbox unchecked.
For example, MailChimp, which has a reputation for

user-centered design, shows that you don’t need to highlight
the entire row. It relies solely on the checked state of the
checkbox. We can assume their research showed this to be
enough. My own research aligns with this too.
An Inbox 231
MailChimp’s campaign list page with one campaign selected.
We could highlight the entire row using CSS and JavaScript,

but we should only do that if user research shows this will
add value (inclusive design principle 7).
Actioning Emails
Letting users select multiple emails is all well and good, but
we’re going to want to facilitate actioning them too. This
form has three actions and, therefore, three submit buttons:
Archive, Delete, and Mark as spam.
232 Chapter 5
<input type="submit" name="archive" value="Archive">

<input type="submit" name="delete" value="Delete">
<input type="submit" name="spam" value="Mark as spam">
<ul class="inbox">...</ul>
The nature of this form and the presence of multiple submit

buttons create several new problems that previous chapters
haven’t had to consider. Let’s discuss each of those now.
THE MULTIPLE SUBMIT BUTTON PROBLEM

Implicit submission lets users submit the form by pressing
Enter when focus is within a field. This convention speeds
up submission without having to move focus to the submit
button. This is especially useful for a single field form such
as a search form.
Having multiple submit buttons with differing actions is

problematic because if the user presses Enter, which action
will be taken? The answer is that browsers will choose the
first button in the document source.
The best solution to this problem is to avoid it; that is, to have
just one action per form. Depending on the design, this may
not be easy, which is unfortunately the case with the inbox.
One alternative approach could be to expect users to choose

which action they want to perform, before selecting the
An Inbox 233
emails to apply that action to. But this seems somewhat

unconventional and long-winded.
Fortunately, multi-select interfaces usually place the submit

buttons at the top of the form in close alignment to the
checkboxes. This gives users a way to discover the available
actions before making their selection.
Gmail’s inbox screen showing a selected email with an additional menu now
available.
It’s worth noting that implicit submission is probably less

useful on a form consisting solely of checkboxes. In any
case, as we have multiple submit buttons, we should put the
least critical action first — in this case, Archive. That way, if
a user happens to submit the form implicitly, they’ll be in
less of a predicament.
Also, we can offer users a way to undo their last action,

which we’ll discuss later.
234 Chapter 5
STICKY MENUS
The menu is placed above the list of emails; as users scroll, it
might disappear off screen. A sticky menu, however, would
stay on-screen as soon as the menu gets to the top edge of
the viewport.
Sticky menu in three states. Left: menu positioned above the content as normal,
before the page is scrolled. Center: the menu (not sticky) rolls off-screen after the
page has been scrolled. Right: a sticky menu still on-screen even after the page
has been scrolled.
Similarly, Google’s material design

has the floating action button. As
users scroll, the action button floats
on top of the content. Both of these
techniques give users quick and
easy access to the menu without
having to scroll back up to the top.
Floating action button layered on top of the

screen at the bottom-right of the viewport.
An Inbox 235
However, sticky menus are problematic for three reasons.
First, they obscure the content beneath, which is especially

distracting on mobile as the menu impedes access to the
primary content. In the case of the inbox, the primary need
is to read and respond to email, not to bulk action it.
Second, sticky menus are usually employed to solve symp-

toms that mask the true underlying problems; for example,
that the page is often too long in the first place. An inbox
typically shows just 20 emails at a time, which means the
menu is, at most, a quick flick away on mobile and always in
view on desktop.
Third, the items within the sticky menus are difficult to

focus with the keyboard. You might be halfway down the
page but the menu (which is in close proximity visually)
could be a long way away via the keyboard.
For these reasons, it’s better to position the menu statically.
Note: Where sticky menus are useful, you can use position:
sticky as a progressive enhancement. In the past, we had
to resort to complicated techniques that created jarring and
broken experiences across a range of mobile and tablet devices.4
4 http://smashed.by/fixedposition
236 Chapter 5
DISABLING AND HIDING BUTTONS

Some multi-select interfaces will hide or disable the menu
buttons until at least one item has been selected. You could
argue that showing (or enabling) the buttons in response
to selecting an item helps users take the next step. When
hiding the buttons, the interface becomes more streamlined
as the buttons are only shown as they become relevant. But
this is problematic for three reasons.
First, hiding the buttons means the available actions aren’t

discoverable. This is why designers opt for disabled buttons.
But we discussed the problems with disabled buttons in
chapter 1, “A Registration Form.” As a quick reminder, they
don’t tell users why they’re disabled, and screen reader users
can’t focus them.
Second, there needs to be space to reveal the buttons in the

first place. When there isn’t, the page can judder as the page
reveals the buttons and moves other parts of the interface
around to make space.
Third, having the buttons appear when clicking a checkbox

is distracting as users are focused on selecting the right
emails. And assuming the change of state is valuable, the
buttons would have to be in the viewport for users to see the
change anyway.
An Inbox 237
Just show the buttons at all times.
A Responsive Menu
When there’s enough space, the buttons should just be laid
out at all times, making them readily available and interac-
tive. But if you have more than three buttons in the menu,
or you need to display additional components along the
same row, it’s going to be hard to fit them on screen, espe-
cially on mobile.
Left: on mobile with menu buttons stacked. Right: on desktop with menu
buttons laid out in a row.
The problem is that the buttons will start to stack beneath

one another, which pushes the main content downward
and changes the spatial relationship between the menu and
the list of emails. Moreover, having the menu dominate
the interface is problematic because dominance is a quality
238 Chapter 5
we should use sparingly. After all, if everything dominates,

nothing does. Really, the inbox itself should take center
stage, with the menu taking a back-seat role.
We can handle this problem by hiding the buttons behind a

menu. There are two ways to create a menu: first, by using a
select box; second, by creating a custom menu component.
Let’s discuss the pros and cons of each next.
A SELECT BOX MENU

Select boxes are a menu of sorts. In fact, sometimes, they’re
referred to as dropdown menus, among other names.
Like a menu, they group similar items together that users

can select. And they hide the items behind a click, keeping
the interface compact. They’re an attractive option because,
as we know, browsers supply them for free. But even though
select boxes look like menus and behave like them, and even
though they are sometimes referred to as menus, they aren’t
true menus.
Select boxes are for input. That’s why forms that contain
select boxes — like any other input — must be accompanied
by a submit button to submit the choice. Not only is this
convention, but it’s also in the Web Content Accessibility
Guidelines (WCAG):5
5 http://smashed.by/constbehavior
An Inbox 239
Changing the setting of any user interface component does not
“ automatically cause a change of context […]
The reason I bring this up is because using a select box as

a menu often causes designers to omit the submit button
from the interface. And then JavaScript is needed to submit
the form when the selected option is changed (onchange).
But this submits the form, without the user’s say-so, which
fails inclusive design principle 4, “Give control.”
There are also problems for screen reader and keyboard

users. For example, on Chrome (Windows), the onchange
event is fired as soon as the user presses Down to select
the next option. But with this approach in place, the form
is immediately submitted, making it impossible to move
through all the items in the menu.
Expanded select box with the first option selected. Pressing down immediate-
ly submits the second option when the user might have wanted to select the
third option.
Other browsers are more forgiving of such techniques —

most won’t fire the onchange event (and thus submit the
240 Chapter 5
form) until the user presses Space or Enter. But not all
browsers are alike nor implement the specification consis-
tently. Ignoring people who use one of the less forgiving
browsers doesn’t make the problem any less real.
The other problem with using a select box is that it’s always
collapsed, even when there’s enough space to lay out the
options. One solution is to use JavaScript to create a com-
pletely different component for big screens. This is known
as adaptive design.6
ADAPTIVE DESIGN VS. RESPONSIVE DESIGN

First a fun history lesson.
When the web came along, we settled on 640 pixel widths (as
computer monitors commonly supported this resolution).
Then a few years later, when larger monitors came to mar-
ket, we increased it to 800 and then 960 pixels. We no longer
cared about people with smaller monitors. We expected users
to maximize their browser window; if they didn’t, they’d get a
horizontal scroll bar, and that would be their problem.
More years passed. The mobile web was born. Or, more
accurately, we could use websites on our phones, which
happen to have small screens. A million devices came out.
6 http://smashed.by/rwdadaptive
An Inbox 241
A million browsers came out. And browsers gave us CSS

media queries. Accordingly, we started to design for a width
of 320 pixels. Why? Because many of us had iPhones, and this
happened to be its width in portrait orientation. The hardcore
among us started designing for portrait and landscape sizes
according to the most popular devices at that time.
Now we also have tablets, desktops, and really big desktop

screens. We can browse on large screen televisions and tiny
watches. If your head is spinning, don’t worry, so is mine.
This is the problem that responsive design solves and adap-
tive design exacerbates.
The difference between responsive and adaptive design is

both subtle and crucial. Both techniques are often based on
viewport width. And both use CSS media queries to change
the interface. But they are really quite different.
Adaptive Design
Not everyone in the industry agrees on the meaning of

adaptive design.7 Some think it means having different
layouts that snap at particular sizes — in other words, not
fluid. Others think it’s the same as responsive design, which
is hardly surprising seeing as the words are synonyms.
7 http://smashed.by/adaptivedesign
242 Chapter 5
The other common understanding of adaptive design is

about defining several different (parts of) layouts, made
up of different HTML that’s rendered. Originally, this was
based on user agent string.8 Sometimes, JavaScript is used to
restructure the arrangement of HTML at certain viewport
widths. But more commonly these days, it’s done using CSS
media queries. We’ll focus on this flavor of adaptive design
from this point.
This involves delivering all the HTML for the different

layouts, and hiding and showing these layouts based on CSS
media queries that match a particular device’s width.


<div class="stuff1">...</div>

<div class="stuff2">...</div>
@media only screen and (min-device-width: 375px) and (max-

device-width : 667px) {
.stuff1 {
display: none;
}
.stuff2 {
display: block;
}
}
8 http://smashed.by/ress
An Inbox 243
This approach is normally unnecessary and counter-

productive for a number of reasons.
• There’s an endless stream of devices and browsers

with different widths: creating specific designs for
every device width is impossible.
• The extra code needed to produce such designs
would result in slow-loading pages, which are detri-
mental to the user experience.
• Not all components need a breakpoint. Plenty of
components can be designed to work well in exact-
ly the same way on both small and large screens.
• More importantly, users should get a consistent
experience (inclusive design principle 3) no matter
which device they choose to use. Rotating a device
from portrait to landscape, for example, shouldn’t
mean having to relearn an interface because that
puts an unnecessary cognitive burden on users.
Responsive Design
Responsive design takes a different approach. It’s about

designing a single, fluid interface that works well at any
size, regardless of device. Specific browsers and device
widths become irrelevant. The difference is that you only
add a media query when and if something breaks. These
media queries are known as content breakpoints.
244 Chapter 5
@media only screen and (min-width: 61.37em) {

/* Fix broken layout for a particular thing at a
particular width here */
}
Where adaptive design tries to bend the web to its will,

responsive design embraces it. Responsive design under-
stands that you can’t possibly design for every device and
browser individually. That’s just not how the web works.
Instead, responsive design encourages us to design inter-
faces that work on any size screen.
The select box design I mentioned earlier requires an

adaptive approach: on small viewports users get a select
box; then, when there’s enough space, it’s swapped out for
submit buttons.
Left: select box menu for small screens. Right: menu buttons laid out in a row
for large screens.
In this case, the big screen view entirely discards the select
box in favour of a different interface using CSS and Java-
Script. We either have to change the HTML dynamically
with JavaScript, or we have to have both layouts in HTML,
ready to be enabled and disabled through a CSS breakpoint.
An Inbox 245
The server also needs to be aware of how both menus trans-

mit data. In this case, the select box and submit buttons
would be sending different data.
Not only is all of this more work, but the page will take
longer to load and there are now two vastly different varia-
tions of the same feature to maintain indefinitely. Adaptive
design should always be a last resort.
HOVER VERSUS CLICK

On the web, menus are sometimes opened on hover. Design-
ers often assume that this aids discovery and saves users
the effort of clicking. The thing is, there are many problems
with opening a menu (or anything really) on hover and the
effort of clicking a part of the interface is extremely low.
First, hovering is not an intention to open the menu. When

a user moves over a hover menu it can obscure the content
behind it, which disrupts the experience. With the inbox, as
the user goes to select the first checkbox, they may acciden-
tally end up clicking one of the items in the menu, which
fails inclusive design principle 4, “Give Control.”
246 Chapter 5
Second, users have to be careful to keep the cursor within the

bounds of the menu, otherwise it will close. This is known
as a hover tunnel, and is especially difficult to operate with
motor impairments.
Third, not all users use a mouse (or other types of pointing
device) and touchscreen devices are usually operated with-
out one.
You should note that opening a menu on hover on desktop

and on click for mobile isn’t recommended either. There
are many large touchscreen devices and many small screen
laptops. Features should never be inferred from screen size.
Needless to say, menus should be triggered on click, which is

an explicit intention to activate it, keeping users in control.
A TRUE MENU
Having explored the pitfalls of adaptive design and hover
menus, we can now safely proceed to design a true respon-
sive menu that opens on click.
An Inbox 247
Left: a collapsible menu for small screens. Right: a menu bar for large screens.
The Basic Markup
<div class="menu">
<div role="menu">
<input type="submit" name="archive" value="Archive"
role="menuitem">
<input type="submit" name="delete" value="Delete"
role="menuitem">
<input type="submit" name="spam" value="Mark as spam"
role="menuitem">
</div>
</div>
Notes
• The menu itself has role="menu" indicating that
it contains menu items. When a menu item is
focused, screen readers will announce it as a three-
item menu.
248 Chapter 5
• The menu role isn’t commonly used in web applica-

tions. Only ones that mimic desktop applications
(like ours, which is essentially a web-based email
application) apply.
• The wrapping <div class="menu"> will be needed
for enhancement purposes because the toggle but-
ton will be prepended to it.
SMALL MODE VERSUS BIG MODE

When the script initializes, it will need to check to see if the
viewport is in small or big mode. We’re using the words small
and big, as opposed to mobile and desktop, because respon-
sive design doesn’t think in terms of devices. Moreover, the
media query values for small and big don’t necessarily cor-
respond to mobile or desktop — they are determined by the
place in which the menu would otherwise break.
The constructor function (shown below), takes two

arguments: the container element and mq (short
for media query) string. The media query string is
(min-width: 45em) because that’s where the interface
starts to break.
function Menu(container, mq) {

this.container = container;
this.menu = this.container.find('[role=menu]');
this.mq = mq;
An Inbox 249
this.keys = { esc: 27, up: 38, down: 40, tab: 9 };

this.menu.on('keydown', '[role=menuitem]', $.proxy(this,
'onButtonKeydown'));
// create button and listen to click and down events
this.createToggleButton();
// Setup up media query listener and check which applies
on initialisation
this.setupResponsiveChecks();
}
Besides assigning properties to this to make them available

to other methods (shown later), the constructor is respon-
sible for listening to the keydown event on the menu items
and creating the toggle button.
The last line calls the setupResponsiveChecks() method,

which is responsible for collapsing the menu items behind a
traditional menu using a combination of CSS media queries
and JavaScript’s matchMedia API.
Menu.prototype.setupResponsiveChecks = function() {
this.mql = window.matchMedia(this.mq);
this.mql.addListener($.proxy(this, 'checkMode'));
this.checkMode(this.mql);
};
Menu.prototype.checkMode = function(mql) {
if(mql.matches) {
this.enableBigMode();
} else {
this.enableSmallMode();
}
};
250 Chapter 5
The matchMedia API is the JavaScript equivalent of

a CSS media query. Where @media() {} is for CSS,
matchMedia() is for JavaScript. It’s a way of keeping
behavior and style in sync, based on the same media query.
In this case, when the (min-width: 45em) media query
is matched, big mode is enabled. When it doesn’t match,
this means the viewport width is less than 45em, and so
the script calls the enableSmallMode() method which
constructs a toggle menu.
<div class="menu">
<button type="button" aria-haspopup="true" aria-
expanded="false">
Actions
<span aria-hidden="true">▾</span>
</button>
<div role="menu">
<input role="menuitem" type="submit" name="archive"
value="Archive">
<input role="menuitem" type="submit" name="delete"
value="Delete">
<input role="menuitem" type="submit" name="spam"
value="Mark as spam">
</div>
</div>
An Inbox 251
Notes
• The aria-haspopup attribute indicates that the
button shows a menu. It acts as warning that, when
pressed, the user will be moved to the pop-up menu.
• The <span> contains the Unicode character for a
down arrow. Conventionally, this indicates vis-
ually what aria-haspopup does non-visually —
that pressing the button reveals something. The
aria-hidden="true" attribute prevents screen
readers from announcing “down pointing triangle”
or similar. Thanks to aria-haspopup, it’s not need-
ed in the non-visual context.
• The aria-expanded attribute tells users whether
the menu is currently expanded (open) or collapsed
(closed) by toggling between true and false
values.
Note: Before matchMedia, we had to use flaky techniques

to get the width of the viewport, breaking the experience
in various browser and device combinations.9 Even
in browsers that returned the correct viewport width,
it would only do so in pixels — not ems. Using ems is
preferred because when the user increases the text size, the
layout will adapt in proportion.
9 http://smashed.by/rwdjs
252 Chapter 5
Keyboard and Focus Behaviour
When the menu button is clicked, the script checks to

see if the menu is currently open by checking whether
aria-expanded is set to false. If it is, the menu is shown,
and focus is moved to the first item; if it isn’t, the menu is
hidden, and focus moves back to the menu button.
Menu.prototype. {
if(this.menuButton.attr('aria-expanded') == 'false') {
this.showMenu();
this.menu.find('input').first().focus();
} else {
this.hideMenu();
this.menuButton.focus();
}
};
We can use the [aria-expanded] CSS attribute selector to

toggle the menu’s display.
[aria-expanded="true"] + [role=menu] {
display: block;
}
[aria-expanded="false"] + [role=menu] {
display: none;
}
When focus is on a menu item, pressing Down or Up

arrows will move to the next or previous item, on loop.
An Inbox 253
Pressing Escape on a menu item will move focus to the

menu button and closes the menu. Sometimes, Home and
End are used to go to the first and last items directly, which
is particularly useful if there are lots of menu items.
Select All
Users may want to archive every email in their inbox.
Rather than selecting each email one by one, we can provide
a more convenient method. One way to service this func-
tionality is through a special checkbox, placed at the top and
in vertical alignment with the other checkboxes, creating a
visual connection. Clicking it would check every checkbox
in one fell swoop.
MailChimp’s campaign list page showing a select all checkbox positioned

top-left of the list.
254 Chapter 5
Arguably, this standard checkbox has all the ingredients

of an accessible control. It’s screen reader and keyboard
accessible. It communicates through its label and change
of state. Its label would be “Select all” and its state would be
announced as “checked” or “unchecked.” All this behavior
without any JavaScript.
By now, the benefits of using standard elements should be

well understood. Despite this control being accessible by
mouse, touch, keyboard, and screen readers, it just doesn’t
quite feel right. Accessibility is only a part of inclusive
design. This control should look like what it does.
The trouble with using a checkbox is that they don’t signal

what they do. Like select boxes, they are associated with col-
lecting data for submission. We should match people’s expec-
tations by using the same interface component for the same
job. In doing so, the interface becomes familiar and consis-
tent which speaks to inclusive design principle 3, “Be consistent.”
Instead, we can employ a simple button, labeled “Select all,”

that when clicked will check all the checkboxes. At the same
time, the button’s label will change to “Deselect all.” Clicking
the button will then uncheck all the checkboxes putting
them back into their original state.
An Inbox 255


<button type="button">Select all</button>

<button type="button">Deselect all</button>
Note: We looked at how to implement an alternative toggle

button using the aria-pressed attribute in chapter 1 for
the password reveal pattern (see page 39).
Success Messages
When the user submits the form, the selected emails will
disappear from the inbox. When an action has been com-
pleted, telling users is the respectful thing to do. Not doing
so leaves users wondering what happened, if anything.
In chapter 1, “A Registration Form,” we designed and con-

structed an error summary panel that resides at the top of
the page. A success message needs a similar treatment with
a couple of tweaks.
Instead of being red, it should be green, which is conven-

tionally associated with success. Second, the content should
be “You’ve successfully archived 15 emails” (or similar).
256 Chapter 5
A green success message panel.
<div class="successMessage" role="alert">

<h2>You’ve successfully archived 15 emails</h2>
</div>
Both the error and success message panels are placed within
the natural flow of the document and toward the top of the
page to indicate their importance. The role="alert" attri-
bute ensures screen readers will announce it when the page
loads or if it is updated on the client.
TOAST MESSAGES
Some applications employ what is known as a “toast” mes-
sage or notification. When the application needs to notify
users, a little (non-modal) dialog will pop up on the page — a
bit like a piece of toast from a toaster. Then, after a certain
amount of time, the notification disappears automatically,
usually with a fading animation.
A toast notification on Windows, positioned bottom-right, just above the taskbar.

An Inbox 257
This is all very interesting from a design perspective, but

it’s hardly a useful way to communicate. First, the message
obscures the content beneath. Second, users have to read
the message before it disappears. This makes comprehen-
sion a stressful task and takes control away from the user.
A success message should be laid out bare and placed within

the natural flow of the page. There’s no need to obscure
parts of the interface. After all, the message is temporary
and will naturally disappear when the user leaves the page.
However, if users are likely to stay on the page for a long

time after, research might show that dismissing a message
is valuable after all. Offer that functionality with a button.
When clicked, it hides the message.
A success message panel with “Dismiss” button.
Be careful to inject the <button> with JavaScript. If we put

it in directly in the HTML, then the interface will appear
broken when JavaScript is unavailable; that is, nothing will
happen when it’s clicked.
258 Chapter 5
CONFIRMING VERSUS UNDOING

As a safety measure, some roads have speed bumps. They
compel drivers to slow down on roads that are more likely
to cause accidents. We can create a digital speed bump
by asking users to confirm their action by asking them if
they’re sure.
An “Are you sure” confirmation screen with an option to confirm or cancel

the action.
This is fine for infrequent tasks, but it quickly becomes

tedious when that action needs to be performed more often.
Continuing with the driving analogy, then: it’s a bit like put-
ting speed bumps on the motorway. They’d probably cause
more accidents than they’d stop.
An alternative approach would be to let users perform the

action immediately, without any warning. Then, along with
the success message, give users the chance to undo the
An Inbox 259
action. Clicking “Undo” would reverse the action by restor-

ing the emails back to the inbox. If only we could undo
accidents on the road.
A success message panel with “Undo” button.
Summary
In this chapter we began by choosing the right way to
present a collection of emails and the impact of combining
two disparate modes — reading email and actioning it —
into one interface.
We then looked at how a multi-select form is different from

most other types of form, and how this caused us to con-
sider several other visual and interactive design treatments.
Finally, we looked at ways in which to add value and put

users firmly in control by designing consistent interfaces
that give users feedback and a way to undo their actions.
260 Chapter 5
THINGS TO AVOID
• Using the wrong element for the job and fixing it

with JavaScript.
• Using ARIA when standard HTML can be used
instead.
• Using checkboxes and select boxes for buttons and
menu components.
• Doing work that doesn’t add value, such as high-
lighting selected rows.
• Disabling submit buttons until the form becomes
valid.
• Hiding notifications automatically.
• Putting speed bumps in front of repetitive tasks.
DEMOS
• Inbox: http://smashed.by/inboxdemo
262 Chapter 6
A Search Form
I
’m an organized person. Even as a boy, I remember
always having a place for things. To be fair, I’ve always
been minimalist too. Organizing when you only own a
few things is easy. So it’s no surprise I rarely lost things. On
the odd occasion I did, I just shouted in the general direc-
tion of the resident search engine: “Where’s my…,” and I’d
have my answer.
By search engine, I mean Mum! Mum knew where every-

thing was, not just my stuff — everyone’s. This was one of
her many excellent qualities. She didn’t just know where
stuff was, she knew the answer to everything (at least, that’s
how I remember it). If I had grown up with search engines, I
might have nicknamed her Google.
As I’ve gotten older and become a husband and father, my

life is richer but also less minimalist. Even if I meticulously
organize all our belongings, it’s still hard to remember
where it all is. Worst case scenario: I have to peruse each
cupboard hoping that the thing hasn’t been lost — which is
time-consuming.
A Search Form 263
In this chapter, we’re going to design a responsive search

form. Like Mum, we’ll want it to be readily available and
on hand to answer any question users have. To make this
happen, there are some crucial things to consider.
Search Everything
Not only was I able to ask Mum where my stuff was, really
I was able to ask her anything. When designing a global
search form, users should be able to do the same thing. Too
often, users can only find stuff that lives in the database.
On Amazon, search will only return products. On YouTube,
search will only return videos.
In “Content and Design Are Inseparable Work Partners”

Jared Spool explains that “content is the thing the user
needs right now.”1 He recounts a story from user research
where someone was trying to buy a purse.
The woman was happy enough to buy the purse on the

proviso she could return it. But the returns policy wasn’t on
the product page, or in the FAQ. In the end, she tried typing
“Refund policy” into the search box — but it didn’t return
any results. That was the end of the research session.
1 http://smashed.by/contentanddesign
264 Chapter 6
The returns policy isn’t a product. Nor does it reside in the

database. But this is what she wanted. We often hear how
content is king, but the design of the search function let
the content down.
Wherever possible, search should search everything. And if it

doesn’t, the label should be explicit. If the search function only
returns products, make that clear within the interface.
Left: generic search label “Search.” Right: specific search label “Search products.”
A tip: use analytics to track what users are searching for.

If the most popular searches retrieve empty results, make
provisions to improve the experience based on data.
The Basic Form

The search form is simple enough and contains just three
elements: the label, search input, and submit button.
A Search Form 265
A search form.
<div role="search">
<form>
<div class="field">
<label for="search">
<span class="field-label">Search</span>
</label>
<input type="search" id="search" name="search">
</div>
<input type="submit" value="Search">
</form>
</div>
Notes
• As noted in chapter 3, “A Flight Booking Form,” the
search input (<input type="search">) lets users
clear the field more conveniently than a standard
text box, by clicking the delete cross or by pressing
Escape.
266 Chapter 6
• The search form has a search landmark role:

role="search". Like other landmarks, this means
it will be listed as a shortcut in most screen readers.
The extra <div> is necessary because putting the
landmark attribute directly on the <form> would
override its semantics.2
There’s No Room
Typically, search is placed within the header. Like naviga-
tion, this makes it easily discoverable and quick to access.
Putting such an integral feature elsewhere on the page
would be counterintuitive and unconventional.
The challenge, of course, is that it’s hard to fit the search

form inside the header along with everything else. The
header is premium screen real estate. That is, there isn’t
much room available and it’s highly sought after. The more
we put into the header, the more the main content is pushed
down the page.
As noted in earlier chapters, we’re often seduced by novel,

space-saving techniques, such as the hamburger menu,3 but
hiding content should always be a last resort. On desktop,
2 http://smashed.by/searchrole
3 http://smashed.by/hamburgermenu
A Search Form 267
the issue of space isn’t much of, well, an issue — there’s usu-
ally plenty of room. On mobile, though, we’re going to have
to think a bit more. There’s only so much space available to
play with.
Let’s look at some ways to reduce the size of the search

form, so that it might fit inside the header more easily. We’ll
discuss various problems and considerations we need to
think about with each technique.
PLACE THE SUBMIT BUTTON INLINE

As explained in chapter 1, “A Registration Form,” the best place
for the submit button is directly below the final field. But one
way to reduce the amount of vertical space the search form
takes up is to place the submit button next to the field.
Left: submit button below the search field. Right: submit button next to the
text box to save vertical space.
268 Chapter 6
It’s a bit of a special case, and it’s acceptable if the search

form consists of just one form field. But some search forms
offer more than one field. In this case, the submit button
should go directly below the last field as usual.
HIDING THE LABEL

Another space-saving technique is to hide the label. You
might consider using the placeholder attribute to supplant
the label, but this is problematic for a number of reasons. See
chapter 1, “A Registration Form,” for an in-depth rundown.
We could forgo a visible label altogether because arguably

the submit button acts as a quasi-label for sighted users. But
you should still include a label for screen reader users: they
shouldn’t have to skip ahead to the button in the hope that
its label provides a clue.
<div class="field">
<label for="search" class="visually-hidden">
<span class="field-label">Search</span>
</label>
<input type="search" id="search" name="search">
</div>
Note: The CSS for the visually hidden class is set out in
chapter 2, “A Checkout Form” (page 116).
A Search Form 269
If search only retrieves products, for example, then the

button would be better labeled “Search products.” This way,
users know what they can and can’t search for. As noted
above, users should be able to search everything — although
technological and resource limitations may come into play.
Also, if you recall the hint and error patterns from “A Registra-
tion Form,” (on page 56) the text is injected into the <label>.
If you need to show this information, you’ll have to come up
with a new solution, which seems unnecessary. Besides, hid-
ing the label doesn’t usually save enough space to fit the form
inside the header anyway, especially in smaller viewports.
HIDING THE BUTTON

We might also consider hiding the button, but this has
several pitfalls. First, without a button you can’t use it as a
quasi label; the label would need to be reinstated, taking up
room again.
Second, the button is a fundamental part of a form — without

it, it’s not clear how users are meant to perform the search.
While we may be aware that forms can be submitted implic-
itly (by pressing Enter), not all users are. See chapter 5, “An
Inbox,” for more information about implicit submission.
Third, if your search form contains more than one field,

omitting the submit button stops implicit submission from
270 Chapter 6
working. Fortunately, if your search form consists of a single

field, implicit submission will still work without a button.
Interestingly, many of the sites that omit the submit button

normally find room to include a magnifying glass icon to
signify its otherwise hidden affordance. In this case, they
may as well place the icon inside a submit button solving
both problems at the same time.
to the left of the text box.
Medium’s search form lacks a submit button but has a magnifying glass icon
Note: If you decide to hide the button using the

visually-hidden class, remember the button will still
be focusable. This means sighted screen reader users, for
example, will find this problematic.4 You can fix this by
adding the tabindex="-1" attribute.
Toggling the Form’s Visibility

Even if removing the label and submit button didn’t degrade
the usability of the form, it still wouldn’t really solve the
space problem enough to fit the form comfortably within
the header.
4 http://smashed.by/notallblind
A Search Form 271
Instead of messing around with pixels to this extent, we

can toggle the entire form’s visibility by letting users click a
button. Finding room in the header for a button is relatively
straightforward. And including a visible label, a hint (if
needed), and a submit button is an easy task.
Left: hidden search form. Right: search form revealed.
THE BASIC MARKUP

The basic page consists of a header containing the logo
and navigation. The search form is positioned underneath
the header.
<header>
...
</header>
<div role="search">...</div>
272 Chapter 6
THE ENHANCED MARKUP

When JavaScript is available, the markup is enhanced
like this:
<header>
...
<button type="button" aria-haspopup="true" aria-
expanded="false">Search form</button>
</header>
<div role="search" class="hidden">...</div>
Notes
• A toggle button is injected into the header.
• The search form is hidden using the hidden class as
introduced in chapter 1.
• The aria-haspopup attribute indicates that the
button reveals a part of the interface (the search
form). It acts as warning that, when pressed, the
focus will be moved to the search form.
• The aria-expanded attribute tells screen reader us-
ers whether the search form is currently expanded
or collapsed by toggling between true and false
values.
A Search Form 273
A SMALL SCRIPT
function SearchForm() {
this.header = $('header');
this.form = $('.searchForm');
this.form.addClass('hidden');
this.button = $('<button type="button" aria-
haspopup="true" aria-expanded="false">
<img src="/public/img/magnifying-glass.png" width="20"
height="20" alt="Search products"></button>');
this.header.append(this.button);
}
SearchForm.prototype. {
if(this.button.attr('aria-expanded') == 'false') {
this.button.attr('aria-expanded', 'true');
this.form.removeClass('hidden');
this.form.find('input').first().focus();
} else {
this.form.addClass('hidden');
this.button.attr('aria-expanded', 'false');
}
};
Notes
• The constructor is responsible for enhancing the
HTML and listening to the button’s click event.
• When the button is clicked, the script checks the
aria-expanded attribute to see if the form is cur-
rently expanded or collapsed.
274 Chapter 6
• If the form is collapsed, then the aria-expanded

attribute is set to true, and the hidden class is
removed to reveal the form. Finally, the first input
is focused, which saves users an unnecessary
extra click.
• If the form is expanded, the aria-expanded
attribute is set to false and the form is hidden
by adding the class of hidden to the form.
Displaying Search Results

Displaying search results is somewhat out of scope for a
book about form design, but let’s run through some import-
ant details quickly now:
• Maintain search text. When the user arrives at

the search page, what they typed should persist.
This way users can make tweaks without having to
retype the entire query.
• Display result count. Tell users how many results
have been returned. A simple approach would be
to update the page’s <title> text to read “Search
results for [search term]” or similar. Users can then
decide what their next action is. For example, if
there are many results, they may decide to filter
them (more on this in the next chapter).
A Search Form 275
• Let users sort. Depending on the dataset being

searched, it’s often useful to let users sort by rele-
vance, popularity, or recency, for example.
• Don’t employ infinite scrolling by default. It’s an
anti-pattern with several usability issues.5 This leaves
“Show more” or standard pagination. “Show more”
is more appropriate for sites with a lot of user-gener-
ated content, where the location of the result is not
important. Pagination is more appropriate for ecom-
merce sites, where users are looking for a specific
item, not just browsing for entertainment.
Summary
In this chapter we started by looking at how important it
is to give users what they searched for — not just products,
or articles, but anything the site contains.
We then went on to look at the interface; specifically, how

we can accommodate a search form as part of the header
so that it’s readily accessible from every page in the site.
We also enhanced the experience for screen reader users

by using the role="search" landmark.
5 http://smashed.by/infinitescrolling
276 Chapter 6
CHECKLIST
• Search everything, not just what’s stored in

a database.
• Use the role="search" landmark to help screen
reader users access search quickly.
• If the form can’t fit easily in the header, let users
toggle its display with JavaScript.
• When displaying search results, avoid infinite
scrolling.
Demos
• Search form: http://smashed.by/searchformdemo
278 Chapter 7
A Filter Form
I
n the introduction to “A Search Form” you’ll recall the
type of conversation I used to have with Mum. Some-
times I would ask, “Where’s my black top?” But this
was so vague that Mum would respond with questions like,
“Is it a football or tennis top?” This question is a filter on
a large set of results. Without knowing the answer to this
question, Mum couldn’t respond with an accurate answer.
Filters (also referred to as facet navigation or guided naviga-

tion) let users refine a large set of search results. This helps
users home in on what they’re looking for.
First of all, though, it must be said that if you don’t need

a filter, don’t include one. They’re only useful if searching
returns a vast amount of results. On Google, where search-
ing can yield thousands, if not millions of results, most
people aren’t willing to click beyond the first or second page.
Letting users filter out irrelevant results is important. The

ability to filter not only offers an additional dimension of
control, but it does so in a way that matches each user’s own
mental model. In “Designing for Faceted Search,” Stephanie
Lemieux says:
A Filter Form 279
Think of a cookbook: authors have to organize the recipes in one
“ way only — by course or by main ingredient — and users have to

work with whatever choice of organizing principle that has been
made, regardless of how that fits their particular style of searching.
An online recipe site using faceted search can allow users to decide
how they’d like to navigate to a specific recipe, e.g. by course type,
cuisine or cooking method.1
At first glance, filters might look similar across different

sites, but their behavior varies quite widely. When and how
filters should be applied, how to denote selected filters,
what elements should be used, how to give users feedback,
how they’ll work on mobile and desktop: all of these things
need to be taken into account.
Interactive Filters versus Batch Filters

There are two ways to let users filter: one at a time (interac-
tive filtering), or selecting multiple filters at once (batch fil-
tering). In “User Intent Affects Filter Design,” Katie Sherwin
describes an excellent way to think about this:2
[…] think about how you might order appetizers at a restaurant.
“ Say you want to order three appetizers for the table, but as soon as
you name the first one, the waiter snatches the menu out of your
1 http://smashed.by/facetedsearch
2 http://smashed.by/applyingfilters
280 Chapter 7
hands and walks back to the kitchen to get the chefs started on
cooking that dish. Instead, a good waiter understands that you’re
still in the process of ordering and knows to give you more time
before taking away the menu. A good waiter allows you time
to make a batch decision, even if that might slightly delay the
delivery of the first item ordered. (However, sometimes the waiter
may take the appetizer order, and then give you more time to decide
on the main course. A good waiter is flexible and adapts to the needs
of the customers.)
INTERACTIVE FILTERS
Interactive filters update as soon as the user clicks a filter.
The advantage is that users will see the results update as
they go.
Left: an interactive filter with no filters selected. Right: the same, but with the
red filter selected.
A Filter Form 281
One disadvantage is that each click causes a page refresh,

which can cause frustration due to lag and getting scrolled
back to the top of the page — something that will happen
every time the user selects a filter. This is especially prob-
lematic for keyboard users as they’ll have to tab back to
where they were.
BATCH FILTERS
Batch filters work by letting users set a number of options
before submitting and reloading the page (see above). One
advantage of this approach is that it’s faster, as users just
make one request for several filters.
Left: a batch filter with filters about to be submitted. Right: the page with the
filters applied.
One disadvantage of this approach is that a combination of

filters could lead to zero results.
282 Chapter 7
MATERIALLY DISHONEST INTERFACES

We’ve already discussed the concept of material honesty
several times in the book. This is because, rather unfortu-
nately, dishonest interfaces are prevalent on the web. As a
reminder, one material shouldn’t be used as a substitute for
another because the end result is deceptive. In short, we
should use the right material for the job. But what does this
mean for filters?
As shown above, interactive filters tend to use links because

they provide the expected behavior — that clicking a filter
would immediately request the new page of results. But
some sites style links to look like checkboxes by using CSS
background images, for example.3 But checkboxes are for
input, not for requesting new pages.
Links styled as checkboxes.
3 http://smashed.by/cssbackground
A Filter Form 283
The problem is that a link should look and behave like a link,
not a checkbox. Batch filters, made from real checkboxes, let
users select several filters. Making links look like check-
boxes means users wouldn’t expect clicking a filter would
immediately request the new results. That’s materially
dishonest and therefore deceptive.
Why would designers do this? One possibility is that with-

out making the links look like checkboxes, users wouldn’t
know that they could choose multiple filters within
the same category. Or, in the case of radio buttons, they
wouldn’t know they could choose only one filter.
Breaking widely understood conventions without a very

good reason can seriously harm the user experience. That’s
all well and good as a theory, but how do we take this infor-
mation and design a filter that works?
We’ll find out by taking things step by step.
WHICH TYPE OF FILTER IS BEST?

Only conducting your own research for your problem can
tell you the answer to that, but we’ll go with batch filters.
This seems prudent because:
• users can choose multiple filters at once creating a

faster experience.
284 Chapter 7
• radio buttons and checkboxes have natural signifi-

ers that indicate that one or multiple filters can be
selected within the same category.
• users can still, if they wish, select just one filter at
a time with a batch filter, which speaks to inclusive
design principle 4, “Give control.”
Layout
Before tackling the complexity of the filter form itself, it’s
important to look at it in the context in which users are
likely to use it. In modular design, we can fall prey to focus-
ing so deeply on the individual components that we forget
to check how everything works when they combine to form
the page (or journey).
The wallets category page as seen on desktop, with filters on the left and
results on the right.
A Filter Form 285
In this case, the page has been carefully arranged so the

relationship between the filter and the products is clear.
This has been achieved by using the correct heading levels:
1. The primary heading (level 1) is at the top: “Wallets.”
2. Then there is a subheading (level 2) for each compo-

nent: “Filter” and “Products.”
We’ve married the correct heading hierarchy semantically (as

we’ll see in the code to follow) with how they’re sized and
positioned visually. That is, the top-level heading comes first
and is the biggest. The second-level headings come after and
are smaller. This is important because not only does the fil-
ter let users find products more quickly, but it also provides
context for the products in view.
The Markup
As the form is made from standard form components, the
code for our filter form should be familiar.
<main>
<h1>Wallets</h1>
<aside class="filter" aria-labelledby="filter-heading">
<h2 id="filter-heading">Filters</h2>
<form role="form" method="get" aria-labelledby="filter-
heading?">
286 Chapter 7
<legend>
<span class="field-legend">Color</span>
</legend>
<div class="field-options">
<label for="color">
<input type="checkbox" name="color"
value="green" id="color">
Green
</label>
</div>
<label for="color1">
<input type="checkbox" name="color"
value="red" id="color1">
Red
</label>
</div>

</div>
</fieldset>
<legend>
<span class="field-legend">Rating</span>
</legend>
<label for="rating">
<input type="radio" name="rating"
value="4" id="rating">
4 stars and up
</label>
</div>
A Filter Form 287
<label for="rating1">
<input type="radio" name="rating"
value="3" id="rating1">
3 stars and up
</label>
</div>

</div>
</fieldset>

<input type="submit" value="Apply filters">
</form>
</aside>
<div class="results">
<h2>Products</h2>

</div>
</main>
Notes
• There are three headings on the page: the top level
(<h1>Wallets</h1>) and a level-two heading for
the filter and products components. Many sites pro-
vide an incomplete and broken heading structure
— for example, by replacing <h2>Products</h2>
with <h1>Wallets</h1>. However, this orphans
the <h2>Filter</h2>, which deceives both sighted
and non-sighted users because users expect that a
second-level heading comes after the first.
288 Chapter 7
• The specification has recently changed to allow

headings inside legend elements. We could, then,
consider marking up the text inside the legends as
<h3>s. This would give screen reader users an alter-
native way to navigate the filter, which is some-
times called multimodality and speaks to inclusive
design principle 5, “Offer choice.”
• Notice the form has a role="form" attribute. This
may seem counterintuitive, but it turns the form
into a landmark, which makes it navigable in screen
readers using shortcuts. Since the basic function-
ality works without JavaScript and triggers a page
refresh, this helps users navigate back to the form
from the top of the document. It also means users
can browse the products and still get back to the
filter component quickly.
• Similarly, the filter is marked up as an <aside>,
which is another type of landmark, typically used
to denote a sidebar. An aside should be tangentially
related to the main content, which suits the filter
component well.
• We’re using the GET method on the form to
rebuild the page from the server without relying
on client-side JavaScript at this stage. For example,
submitting the form with the “Red” and “3 stars
and above” options selected will build a page with
A Filter Form 289
?color=red&rating=3 as the query parameter. We’ll

look at enhancing the page with Ajax later.
• The form contains a number of fields which have
been included for the purpose of example. The type
of form control you use should be based on the type
of behavior your users need. As noted in previous
chapters, checkboxes should be used for multiple
selections; radio buttons if only one can be selected.
• Keyboard users can press the arrow keys (left
and right, up and down) to select a radio button.
Pressing the down arrow key, for instance, will
focus and select the next radio button with the
name="rating" attribute: “3 stars and up.” In screen
readers this will announce “rating, three stars and
up, selected, radio button, two of four” (or similar).
Automatic Submission
As noted above, the filter form (like any other form, I
might add) lets users select as many filters as they like
before submitting them. This standard and conventional
behavior should be familiar to users — except this is not
always the case.
290 Chapter 7
I interviewed Dave House, a former designer for Gumtree,

a site which uses filters extensively. Dave and his team
conducted many rounds of usability tests and here’s what
he said:
On desktop, Gumtree users would select filters without submitting
“ them. They didn’t expect to have to submit their choices. We heard a

lot of feedback saying “Your filters are broken.”
Owing to the materially dishonest design of filters as

explained earlier, it seems some people have come to expect
that clicking a checkbox (or radio button) will reload the
results without having to submit. This phenomenon is
known as Jakob’s law:4
Users spend most of their time on other sites. This means that users
“ prefer your site to work the same way as all the other sites they
already know.
Automatically submitting the form when a filter is selected

would effectively convert our batch filter into an interactive
one. That’s a shame, as not only would we be exacerbating
the problem of dishonest design, but we’d be forgoing the
inherent advantages of batch filters.
4 http://smashed.by/endofwebdesign
A Filter Form 291
And this may work for radio buttons and checkboxes, but
what if there were text boxes that could be used to enter a
price range? When would users expect the form to submit?
Submitting while typing is out of the question. This leaves
submitting the form on blur (tabbing or clicking out of the
field), which is odd and unintuitive. We’d need a submit
button just for that box.
If your users have the same expectations as Gumtree’s

users, then you may have no choice. But before going to
such lengths, let’s explore some other techniques to help
users realize that submission is necessary.
First, the button should look like a button and be styled

prominently to stand out on the page. And second, the
button should be within easy reach (within reason). Some
options include:
• Making each filter category collapsible (we’ll look at

how to do this later)
• Duplicate the submit button at the top of the form
• Consider making the buttons stay on screen by using
position: sticky.5
5 http://smashed.by/positionsticky
292 Chapter 7
Should We Just Change to Links?

Perhaps we should just throw away the form and use links.
The advantage would be we wouldn’t need JavaScript —
users would get standard (link) behavior for free.
But there are several disadvantages:
• We can’t use links for the dynamic price range

inputs, for example, which is unnecessarily con-
straining.
• We may need to signify the “select one” or “select
multiple” behavior across different link filters,
which is challenging without making links look like
checkboxes (here we go again).
• If users select many filters then they may exceed
the max limit of the query string. Posting a form
averts this problem.
Really, we’re exchanging one set of problems for another,

arguably larger set. Let’s see how we can submit the form
automatically using JavaScript and the issues that may arise
from doing so.
A Filter Form 293
SUBMITTING THE FORM AUTOMATICALLY

If, despite our efforts to thwart breaking convention, users
still expect the form to submit automatically, we can use
JavaScript to submit the form in response to the form’s
change event.
As our filter only contains checkboxes and radio buttons,

we can first remove the now redundant submit button.
However, we can’t completely remove the button from the
document (with display: none, for example) because
some platforms (iOS for one) will not submit forms when
a submit button isn’t present. And, as mentioned in chap-
ter 6, omitting the submit button stops users being able to
submit the form implicitly, with the Enter key. In which
case, we can use our special visually-hidden class, plus
tabindex="-1" to make sure the button isn’t user-focusable.
function FilterRequester() {
this.form = $('.filter form');
this.form.find('[type=submit]')
.addClass('visually-hidden')
.attr('tabindex', '-1');
}
Now the submit button has been hidden accessibly, we can

submit the form when a filter is changed by listening to the
change event:
294 Chapter 7
//...
this.form.find('[type=radio], [type=checkbox]').
on('change', $.proxy(this, 'onInputChange'));
}
FilterRequester.prototype. {
this.form.submit();
};
You should note that this fails Web Content Accessibility

Guidelines Success Criterion 3.2.2:6
Changing the setting of any user interface component does not
“ automatically cause a change of context
Additionally, keyboard users operating the filters must use

their arrow keys to move through the radio buttons. Each
arrow keypress not only focuses adjacent radio buttons,
but also selects them. As a result, keyboard users won’t be
able to move from one radio button to another without
the form being submitted. What if they want the third or
fourth radio button?
Even if you ignore the difficulties associated with certain

interaction modalities, having the page refresh in the mid-
dle of choosing filters is a poor user experience. Let’s see if
Ajax can fix these issues.
6 http://smashed.by/constbehavior
A Filter Form 295
Ajax
Ajax is a technology that lets users dynamically update
parts of an interface without a page refresh. The advantage
for our filter form is that users can select as many filters as
they like without being interrupted by a page refresh and
the focus moving to the top of the document.
To do this, we can listen to the change event and fire off an

Ajax request:
var data = this.form.serialize();
this.requestResults(data);
};
FilterRequester.prototype.requestResults = function(data)
{
$.ajax({ data: data, ...});
}
While the main issue has been solved, the introduction of

Ajax has created additional problems. Let’s discuss each of
these now and see how we might deal with them.
COMMUNICATING LOADING STATES

When a web page is loading, the web browser shows a load-
ing indicator. This loading indicator is accurate, accessible,
and as it’s part of the browser shell it appears in the same
296 Chapter 7
place no matter the website, making it trustworthy and

familiar.
When Ajax is used, we have to

provide our own mechanism to
inform users that the request
is loading. This is normally the
purview of a loading spinner. A loading spinner.
But you should note that unlike the browser, it doesn’t tell
users how long is left, or if the connection is slow. In the
next chapter, we’ll look at ways to provide an accurate prog-
ress bar with Ajax.
Also, the loading spinner, in its current form, is only deter-

minable by sighted users. To provide a comparable experi-
ence (inclusive design principle 1) for screen reader users, we’ll
employ a live region (as first set out in “A Checkout Form”).
<div aria-live="assertive" role="alert"

class="visually-hidden">Loading products.</div>
When the products are loaded:
<div aria-live="assertive" role="alert" class="visually-

hidden">Loading complete. 13 products listed.</div>
A Filter Form 297
Note: As the loading spinner is enough communication

for sighted users, the live region is given the special
visually-hidden class as set out in “A Checkout Form.”
BREAKING THE BACK BUTTON

When a user loads a web page, the browser refreshes and
puts the previous page into its history. This allows users to
press the back button to return to it quickly. But when Ajax
is used to make updates, it’s not put into the browser’s his-
tory: technically, the user hasn’t navigated to another page.
Despite this, if the page looks like it has significantly

changed, then users will expect the back button to work as
normal. Baymard’s usability study found that breaking the
back button’s behavior caused confusion, disappointment,
anger, and even abandonment.7
This is the sort of thing that can happen when a little

enhancement breaks convention. This is why the best expe-
riences are usually the simplest.
Fortunately, we can use HTML5’s History API, which was

designed to help solve this problem.8 First, we have to create
a history entry when the Ajax request succeeds, like this:
7 http://smashed.by/macysfilter
8 http://smashed.by/historyapi
298 Chapter 7
var data = this.form.serialize(); // color=red&rating=3
this.requestResults(data);
history.pushState(data, null, '/path/to/?'+data);
};
FilterRequester.prototype.requestResults = function(query)
{
$.ajax({ ..., success: $.proxy(this, 'onRequestSuccess',
query)});
};
FilterRequester.prototype. > function(query, response) {
history.pushState(response, null, '/path/to/?'+query);
//...
};
Notes
• The first parameter is the state which we want to
be stored with the history entry. In this case, it’s the
JSON response that’s used to render the updated
page.
• The second parameter is the title. As it’s not well
supported and it’s not necessary in our case, we’re
ignoring it by passing null.
• The third parameter is the history’s URL, which is
the URL including the query string.
A Filter Form 299
With this in place, we need to listen to when the history

changes:
//...
$(window).on('popstate', $.proxy(this, 'onPopState'));
}
this.requestResults(e.originalEvent.state);
};
Notes
• The state property contains the JSON response we
associated with the history entry on creation. It’s
then passed to the already written requestResults
method so it can be used to render the page again
without an AJAX call.
• As we’re using jQuery to listen to the onpopstate
event, the state (normally e.state) property is
found in e.originalEvent.state.
Users May Not Notice the Results Update

Another problem with making updates using Ajax is ensur-
ing that users notice the results update. Take a situation
where the filter component is very long and scrolls beyond
the fold. Imagine that while selecting a filter toward the
300 Chapter 7
bottom, it returns too few results such that the user sees a
blank screen.
Page showing a long list of filters with the few results now offscreen.
We can’t move focus as that defeats the entire point of intro-

ducing Ajax in the first place. There are two ways we might
solve this problem.
First, we can set a maximum height on the filter and give it

an additional inner scroll bar. However, inline scroll areas
are really hard to use; so much so, that Baymard Institute
says they should be avoided.9
Second, we can use progressive disclosure to collapse the

filter categories. We’ll be looking at this in detail shortly.
9 http://smashed.by/inlinescroll
A Filter Form 301
BATTERY DRAIN AND DATA ALLOWANCE

As we’re now making heavy use of Ajax to rerender the page
every time the user selects a filter, this will cause users’ data
and battery to be eaten up at a more rapid rate.
AJAX IS NOT NECESSARILY FASTER

Despite popular belief, Ajax is not necessarily faster than a
page refresh.
First, it must be said, that Ajax is subjected to much of the

same latency as a standard page request. If the site (or con-
nection) is slow, then so too will be the Ajax request. In fact,
an Ajax request can be even slower.
This is mostly because it engineers away progressive ren-

dering (also called chunking) which the browser provides
for free. In “Fun hacks for faster content,” Jake Archibald
explains the concept of progressive rendering:10
When you load a page, the browser takes a network stream and
“ pipes it to the HTML parser, and the HTML parser is piped to the
document. This means the page can render progressively as it’s
downloading. The page may be 100k, but it can render useful con-
tent after only 20k is received.
10 http://smashed.by/fastercontent
302 Chapter 7
As Ajax has to wait for the entire 100Kb before showing any-
thing, users have to wait a lot longer to see something.
This is not to say Ajax is bad. It’s just that we should use it
judiciously and when we know that users will benefit from
it. We introduced Ajax on the assumption that users needed
it, but where possible we should, generally speaking, reserve
the use of Ajax for making smaller page updates, for which
it is better suited.
In the end, we can only be sure of what’s best by conduct-

ing user research with a diverse group of people, using a
broad range of browsers and devices, on varying connection
speeds in the context of our own problem.
Note: Screen size shouldn’t be used to infer fast or slow

connection speeds. Like many other people, most of my
time on mobile involves me being connected to Wi-Fi. An
inclusive experience is one that is made fast for all, by
not adding superfluous features and bloat to a page in the
first place.
Collapsible Filters
If your filter has many categories and many options within
those categories, we need to be sure users aren’t overloaded
with too much choice (as explained in chapter 4, “A Login
A Filter Form 303
Form”). The best way to do this is to have fewer filters, so

don’t include ones that users don’t need.
Additionally, we can collapse the categories. This is advanta-

geous for a number of reasons:
1. The submit button is more likely to draw users’

attention as it will be in view.
2. Ajax-injected results will likely be in view.
3. Users shouldn’t have to scroll nearly as much, while

still being able to scan the filter categories.
4. Keyboard users won’t have to tab through all the

filters to get to the one they want. This is because
hidden content isn’t focusable.
ENHANCING THE MARKUP

The basic markup consists of standard form fields. As an
example, this is what the color field markup looks like:
<legend>
<span class="field-legend">Color</span>
</legend>

</div>
</fieldset>
304 Chapter 7
The JavaScript-enhanced markup will look like this:
<legend>
<button type="button" aria-expanded="false">Color</button>
</legend>
<div class="field-options hidden">

</div>
</fieldset>
The legend now contains a button element with a

type="button" attribute, which stops it from submitting
the form. We don’t want it do that: we just want it to expand
and collapse the filters.
Had we instead converted the legend into a button using

ARIA’s role="button", we would be overriding the legend’s
semantics. This means screen readers wouldn’t announce
the legend as the group’s accessible label. We would have
also had to recreate all the free browser-provided behavior
associated with the <button> element, such as being focus-
able and activated by pressing Space and Enter keys.
STATE
The checkboxes are hidden by the hidden class, as first
explained in chapter 1, “A Registration Form.” Removing it
will reveal the checkboxes.
A Filter Form 305
The button has an aria-expanded attribute, initially set

to false, which denotes that the section is collapsed. When
the button is clicked this will be switched to true, which
means it’s expanded. For screen readers, the button will be
announced as “Color, collapsed, button” (or similar, depend-
ing on the screen reader).
Left: collapsed filters. Right: expanded filters.
We also need to communicate the state of the component

visually. Replacing the entire legend with a button element
is not ideal because we still want the legend to look like
what it is: a legend. By the same token, the interface needs
to make it clear that clicking the legend will toggle the filter.
We can signify this functionality with the conventional plus

(can be expanded) and minus (can be collapsed) symbols,
though up and down triangles may work just as well. Let’s
make use of a lightweight SVG icon placed inside the button:
306 Chapter 7
<button type="Button" aria-expanded="false">

Color
<svg viewBox="0 0 10 10" aria-hidden="true"
focusable="false">
<rect class="vert" height="8" width="2" y="1" x="4" />
<rect height="2" width="8" y="4" x="1" />
</svg>
</button>
The aria-hidden="true" attribute hides the icon

from screen readers — the button’s text is enough. The
focusable="false" attribute fixes the issue that in
Internet Explorer SVG elements are focusable. And the
class="vert" attribute on the vertical line allows us to
show and hide it based on the state using CSS, like this:
[aria-expanded="true"] .vert {
display: none;
}
Script
All the script does is create and inject a button and toggle
visibility when clicked. Here’s the entire script:
function FilterCollapser(fieldset) {
this.fieldset = fieldset;
this.options = this.fieldset.find('.field-options');
this.legend = this.fieldset.find('legend');
A Filter Form 307
this.createButton();
this.hide();
}
FilterCollapser.prototype.createButton = function() {
this.button = $('<button type="button" aria-
expanded="true">'+this.legend.text()+'<svg viewBox="0 0 10
10" aria-hidden="true" focusable="false"><rect class="vert"
height="8" width="2" y="1" x="4" /> <rect height="2"
width="8" y="4" x="1" /></svg></button>');
this.legend.html(this.button);
};
FilterCollapser.prototype. {
this[this.button.attr('aria-expanded') == 'true' ? 'hide'
: 'show']();
};
FilterCollapser.prototype.hide = function() {
this.button.attr('aria-expanded', 'false');
this.options.addClass('hidden');
};
FilterCollapser.prototype.show = function() {
this.button.attr('aria-expanded', 'true');
this.options.removeClass('hidden');
};
Small-Screen Experience
Up to now, we’ve only considered the interface in the con-
text of desktop-sized screens, where there’s enough space to
fit the filter next to the results. But what about the small-
screen experience?
308 Chapter 7
With a mobile-first mindset, if you cut out the superfluous

content and lay out what remains, the experience usually
works well. And this approach scales up easily for large
viewports: increasing the font size and white space is usu-
ally enough.
However, the two components (the filter and the results) are
closely weighted in terms of importance. Really, the filter
needs to be as prominent as the results, something we’ve
been able to achieve in desktop-sized screens.
We can’t just put the filters first, as this will push the results
down the page. And we can’t just put them after the results,
as users would have to move beyond them — most users
wouldn’t know they exist.
We’re left with having to collapse the filters behind a toggle

button. We’ve covered this behavior extensively in chapters 5
and 6. Now we’re going to focus on another related problem.
Earlier, I mentioned part of the interview I had with Dave

House, a former Gumtree designer, who conducted a lot of
research on how to deal with filters. In particular, that desk-
top users expected the filter to update the results as the user
clicked filters with Ajax. However, Gumtree’s research also
showed that on mobile this wasn’t desirable because users
couldn’t see the results refresh. Here’s what Dave said:
A Filter Form 309
On mobile, Ajax wasn’t desirable because users couldn’t see any
“ visible refresh of the results. The amount of filters available in some

categories meant this would have happened off-screen. To the user
this would have looked like nothing had changed. We didn’t want
to move focus to the results on every interaction because users often
wanted to pick more than one filter. We reluctantly had to use an
adaptive approach. On mobile, clicking the “Refine” menu button
would reveal a fullscreen filter panel where users could build up
their refinements and submit when they were done.
Gumtree reluctantly went for an adaptive approach, giving

mobile users and desktop users different experiences. But
there might be a way to give users a more responsive expe-
rience — one that conforms to inclusive design principle 3, “Be
consistent.”
By applying this principle we ensure that users get the

same, familiar conventions applied consistently. In this case,
users learn how to use the filters once, to then use them on
any device or screen size.
We need to make sure that users can see the results update
as they filter. We can achieve this by having the filters
appear on top of the results without completely covering
them. It works because users can see the results on the left,
while filters are selected on the right.
310 Chapter 7
Both Amazon (left) and eBay (right) have the filter appear on top of the
results, allowing users to see the results update as filters are selected.
I found this technique described in “Mobile Faceted Search

with a Tray: New and Improved Design Pattern” by Kathryn
Whitenton, which is worth reading in full.11
Denoting Selected Filters

When the user selects a filter, it becomes checked as stan-
dard. This is an important signifier. Keeping selected filters
in their original position alongside other selected filters is
useful, because some users will remember where the filter
was when they originally selected it.
11 http://smashed.by/mobilefacetedsearch
A Filter Form 311
We can also help users by grouping the selected filter in

a separate list at the top. This confirms to users that what
they selected is being viewed. It also gives users easy
access should they want to remove any of the active filters,
without having to scroll through them scanning for the
marked filters.
We can achieve this by adding a little component to the top

of the filter, like this:
<div class="current-filters" aria-labelledby="current-

filters">
<h3 id="current-filters">Current filters</h3>
<ul>
<li><a href="...">Red (Remove)</a></li>
</ul>
</div>
Notes
• The section is appropriately labeled with a third-
level (<h3>) heading as it sits under the second-
level “Filter” heading.
• The surrounding div is labeled by the heading.
This means the heading will be announced for
screen reader users who have tabbed to (focusable)
links.
312 Chapter 7
Summary
In this chapter we’ve nimbly covered several design details
that often crop up with filters. While we’ve persistently
tried to keep to convention, non-conventional approaches
have been explored that may be needed to satisfy users’
new expectations — expectations that have, unfortunately,
been born out of the many materially dishonest interfaces
present on the web today.
With that said, we’ve carefully made the effort to include a

number of provisions that give users a good and inclusive
user experience, should they need the Ajax-driven, automat-
ically submitted and responsive filter component. And we’ve
done that, by applying five of the seven inclusive design
principles set out in the introduction; namely: provide a
comparable experience, consider situation, be consistent,
give control, and offer choice.
A Filter Form 313
THINGS TO AVOID
• Making links look like radio buttons and checkboxes.

• Automatically submitting forms without exhaust-
ing other simpler techniques.
• Assuming Ajax always delivers a faster and better
user experience.
• Prioritizing best practice above user needs.
Demos
• Filter Form: http://smashed.by/filterformdemo
314 Chapter 8
An Upload Form
The web is more than just text. Whether it’s sending a CV
to a recruiter by email, or adding photos to an eBay advert,
we need to let users upload files. Forms have this capability
baked in, of course.
On one hand, uploading a file is only marginally more com-

plex than, say, inputting text or clicking a checkbox. On the
other hand, there are number of unique design challenges
and opportunities that arise, especially when there’s a need
to upload multiple files at the same time.
As usual, we’ll start by looking at what browsers give us for

free. After that, we’ll look at adding various enhancements
and the various issues that surface as a result of those
enhancements. We’ll end up with a number of different ways
to upload a file, appropriate for several different occasions.
A File Picker
A file picker (<input type="file">) is another type of
form control. When clicked, it will spawn a dialog that lets
users browse files on their computer or device. Once a file is
selected, the dialog closes and the picker updates to reflect
the file has been chosen.
An Upload Form 315
Left: a file picker without a file selected. Right: a file picker with a file selected.
If all users need to do is upload a single file, then you can

add a file picker to your form, and you’re pretty much done:
<form enctype="multipart/form-data">
<div class="field">
<label for="documents">
<span class="label">Choose file</span>
</label>
<input class="field-file" type="file" id="file" name="file">
</div>
<input type="submit" value="Upload" name="upload">
</form>
Notes
• The form has an enctype="multipart/form-data"

attribute, which ensures the file is transmitted to
the server for uploading.
• The file picker uses the same pattern as first de-
scribed in “A Registration Form” and throughout the
book, which can take a hint and error message.
316 Chapter 8
RESTYLING THE FILE PICKER IS DANGEROUS

TERRITORY
Some designers like to restyle the file picker to:
• achieve consistency between different browsers

and operating systems
• match the brand’s look and feel
• be able to configure the control’s text
Whether you agree with all of these reasons or not, it must

be said that pretty and useless is considerably worse than ugly
and useful. But this doesn’t mean aesthetics aren’t important:
where possible we should marry form and function together.
However, it’s just as important to make sure that any tech-

niques we employ to achieve these goals don’t cause any
adverse usability issues. That’s a bit like taking one pill to
fix one symptom, only to need additional pills to relieve the
side effects that come from the first.
Styling file pickers has always been tricky because browsers

ignore any attempt at doing so with CSS. We have to resort
to hacking, which is not usually a good idea — that’s why
it’s called hacking. But let’s walk through how it could work,
what can be achieved, and the pitfalls that are involved.
An Upload Form 317
Hiding the Input
The most robust way of styling the file picker is to visually

hide it, like this:
<div class="field">
<label for="file">
<span class="label">Choose file</span>
</label>
<input class="visually-hidden" type="file" id="file"
name="file">
</div>
Note: The CSS for the visually-hidden class is set out in

“A Checkout Form.”
Now that it’s hidden, we can style the control’s label, which
is easy to style. As described in “A Registration Form,” this
works because a control’s label acts as a proxy to the control
itself: clicking the label is like clicking the input.
Styling the Label
Now the input is hidden, we need to style the label so it

looks interactive. We need to style it as a button and change
its text to “Upload file.”
318 Chapter 8
Left: a label styled as normal. Right: the modified label styled as a button.
Focus States
Now the label looks and is clickable, we need to think about

focus states.
As the input is visually hidden, the user won’t get any feed-
back that it’s in focus when they tab to it. To do this, we can
use JavaScript to add a class of focused to the label when
the input is focused, which will allow us to style it:
.focused {
/* focus styles */
}
Reflecting the Chosen File
When the user selects a file from the dialog, it’s the input
that will change state (as shown earlier). To reflect the cho-
sen file, we’ll need to update the label text when the input’s
onchange event fires.
An Upload Form 319
$('[type=file]').on('change', function(e) {
// change label
});
Left: the button-styled label before file selection. Right: after selection.
Pitfalls
On the face of it, this implementation is visually pleasing

and still accessible. Keyboard, mouse, and touch users can
operate it normally, and screen readers will announce the
value of the input.
But that’s not all it takes to design a fully inclusive and

robust custom file picker interface. There are a number of
additional considerations that this solution doesn’t solve
very well at all.
1. Updating the label to reflect the input’s value is

confusing because the label should describe the
input and remain unchanged. In this case, screen
reader users will hear “cv.doc” as opposed to “Attach
document.”
320 Chapter 8
2. The interface doesn’t fit with the established con-

vention of providing hint and error text, as set out
in “A Registration Form.” Not only would we need
to think of another way to provide this information,
but it creates an inconsistent and unfamiliar user
experience.
3. File inputs are actually drop zones, which means

they let users drag and drop files (instead of going
through the dialog). Hiding the input means forgo-
ing this behavior, which some users may prefer.
4. There’s not much room inside the button for a large

file name. Remember, good design adapts well to
varying lengths of content.
Considering the pitfalls, the improvement to aesthetics

doesn’t seem to justify the downgrade in usability and utility.
A Multiple File Picker

Very few tasks on the web, require a user to upload just
a single file at a time. Take the two examples from the
introduction to this chapter. Both attaching files to an email
and uploading photos to an eBay advert involve uploading
several files in one go.
The easiest but most problematic way to solve this would be

to add the multiple boolean attribute to the file input:
An Upload Form 321
<input type="file" multiple>

This provides the same method of file selection as described
above, except users can select multiple files from within the
dialog.
Multiple file dialog on macOS.
That would be it, if you ignored two significant problems.
First, users can only select files within a single folder. If

they need to upload files from different folders, they can’t.
Of course, users could move all the files into a single folder
beforehand but this puts the onus on the user.
Second, not all browsers support the multiple attribute.

And when support is lacking, a single file input may be
found wanting.
For example, take a form which asks users to submit

receipts. When the multiple attribute is supported, users
can upload all the relevant receipts and submit them. With-
out support, users can only upload a single receipt.
322 Chapter 8
A multiple file form.
One way to solve this problem involves giving users a way

to add more files as part of the flow:
Multiple file form with extra screen to let users continue adding files.
Not only does this design let users upload multiple files in
unsupported browsers, but it also lets the user review their
submission, which is a useful addition regardless.
An Upload Form 323
A Persistent Upload Form

Most forms are ephemeral — users submit a form and
they’re taken to another page without a form. For example,
when registering, users are taken to a confirmation screen.
But the task of uploading files means it’s useful to give

users a form that persists until they’re finished on their own
terms. This is the persistent form pattern in action.
HOW IT MIGHT LOOK

The user can choose and upload a file repeatedly until
they’ve uploaded all the desired files. At which point, they
can click the Continue button.
Left: a persistent upload form before a file has been uploaded. Right: the same
form with a file uploaded and the upload form beneath.
324 Chapter 8
THE MARKUP
<form enctype="multipart/form-data">
<div class="field">
<label for="documents">
<span class="label">Attach file</span>
</label>
<input class="field-file" type="file" id="documents"
name="documents" multiple>
</div>
<input type="submit" value="Upload" name="upload">
</form>
Note the file input has the multiple attribute. When

used in conjunction with a persistent form, the multiple
attribute becomes a robust enhancement. Where supported,
users can select multiple files at a time, meaning fewer
requests and a streamlined experience.
However, when not supported, users can keep uploading a

single file at a time, as many times as they need to until the
task is finished. This solves the problem I mentioned earlier
regarding uploading files from different folders.
A Drag-and-Drop Enhancement
As noted earlier, the native file input acts as a drop zone to
let users drag and drop files. However, there are two prob-
lems with it.
An Upload Form 325
First, it’s not immediately obvious that dragging and

dropping is even possible — there are no signifiers that
make this behavior perceivable. Second, the drop zone has
a small hit area, which makes it hard to use, especially for
motor-impaired users.
To solve these issues, we’re going to take our persistent

upload form and progressively enhance it with better drag-
and-drop functionality.
WARNING: IS DRAG-AND-DROP NECESSARY?

Depending on the situation, the humble file picker may be
all that users need to upload files. In this case, you may not
need to worry about adding a drag-and-drop enhancement
at all. This way, there’s less code to send to the user. As a
result, page load times are faster, and the interface is simpli-
fied at the same time.
It’s also worth noting that a drag-and-drop enhancement is

just that — an enhancement. It should be used in conjunc-
tion with a standard file picker. First, because users can’t
actually drag and drop files on mobile, for example. Second,
users with dexterity problems, such as tremors, may have
difficulty dragging a file.
By giving users both choices, we’re safely following inclusive

design principle 5, “Offer choice.”
326 Chapter 8
HOW IT MIGHT LOOK

The large drop zone is more ergonomic, especially for peo-
ple with motor impairments. It’s conventionally styled with
a dashed border. However, if your users aren’t familiar with
this convention, you can add instructional text.
Inside the drop zone sits

a button. When clicked, it
triggers the dialog as nor-
mal. The button is actually
a label styled as a button
using the ill-advised tech-
nique from earlier. But I
haven’t gone mad, there’s An upload form enhanced with large
good reason for this. drop zone.
WHY STYLE THE LABEL AS A BUTTON

The drop zone has two methods of interaction: dropping
files onto the drop zone, and clicking the button. Browsers
don’t let you programmatically update a file input’s value
owing to security reasons.1 Because of this, we can’t update
the file input’s value, for example, when the user drops files
onto the drop zone. Therefore, files will be uploaded immedi-
ately with Ajax (which we’ll cover shortly).
1 http://smashed.by/dragdropupload
An Upload Form 327
Remember, the form has two methods of interaction:

dropping files onto the drop zone, and clicking the button.
For consistency we want both approaches to upload files
immediately (drop zone ondrop and input onchange). This
way, users don’t have to think about when (or when not) to
submit the form — that interaction is no longer an option.
THE ENHANCED MARKUP

Here’s the JavaScript-enhanced markup:
<form class="dropzone">
<div class="field">
<label for="files">Upload file</label>
<input type="file" name="files" id="files" multiple>
</div>
</form>
Notes
• The button has been removed because the files will
be uploaded with Ajax onchange.
• The “dropzone” class exists as a way to target this
particular form for enhancement.
328 Chapter 8
DRAGOVER AND DRAGLEAVE EVENTS

When the user is dragging files over the drop zone, they should
be given feedback so they know that the files can be dropped.
Left: drop zone. Right: drop zone while file is being dragged over it.
We can achieve this by adding a class to the drop zone when

the ondragover event is fired. Similarly, we need to remove
the class when the user leaves the drop zone (ondragleave):
Dropzone.prototype. {
e.preventDefault();
this.dropzone.addClass('dropzone-dragover');
};
this.dropzone.removeClass('dropzone-dragover');
};
.dropzone-dragover {
/* styles here */
}
An Upload Form 329
Notes
• e.preventDefault() is called to allow the file

to be dropped onto the drop zone. Without doing
this, the browser will try and load the dropped file
instead.
• We can’t just use :hover because the feedback
should only be given when a user is dragging a
file over the drop zone — not just when the cursor
happens to be over the drop zone.
DROPPING FILES
Next we need to handle the file drop, which we can do by
listening to the ondrop event.
e.preventDefault();
this.dropzone.removeClass('dropzone-dragover');
$('.fileList').removeClass('hidden');
this.uploadFiles(e.originalEvent.dataTransfer.files);
};
Notes
• e.preventDefault() is called to allow the file to
be dropped onto the drop zone. Without this, the
browser will attempt to load the file instead.
330 Chapter 8
• The dragover highlight is removed as the file has

now been dropped.
• The file list component is revealed, ready to give
users feedback as the files are uploaded. More on
this shortly.
• The event object (e) contains information about
the files, which is handed over to the uploadFiles
method (shown below)
Dropzone.prototype.uploadFiles = function(files) {
for(var i = 0; i < files.length; i++) {
this.uploadFile(files[i]);
}
};
This method loops through each file and calls the

uploadFile method, which is explained next.
UPLOADING THE FILE

Uploading the file involves two steps: creating the data to be
sent, and actually sending it.
Dropzone.prototype.uploadFile = function(file) {
var formData = new FormData();
formData.append('documents', file);
$.ajax({
data: formData
url: '/ajax-upload',
An Upload Form 331
type: 'post',
processData: false,
contentType: false
});
};
The FormData API is designed to construct key/value pairs

that represent form fields and their values, which can then
be sent with Ajax, including forms that contain files (like
ours does). First, we create a new instance, then we append
the file data to it.
For convenience, we’re using jQuerys $.ajax method.

Here’s a rundown of the properties used:
Property Description
data The data constructed with FormData.
type Set to “post” because data is being sent.
url The URL/endpoint for which the server will
process the uploaded files.
processData Set to false, which tells jQuery not to
convert the data into a query string. This is
important as we’re sending files, not just text.
contentType Set to false, which tells jQuery not to
override the automatically created header
appropriate for sending files. 2
2 http://smashed.by/multipartform
332 Chapter 8
FEEDBACK
It’s all well and good having uploaded the files to the server,
but at this moment the user hasn’t been given any feedback
as to what’s happened. Perhaps the file couldn’t be uploaded,
for example. There are a number of times we need to give
users feedback: during upload, on success, and on error.
Progress
Files can take a long time to upload, especially if the con-

nection is slow. It’s important to give users feedback during
upload — not just on completion.
We can show feedback with a progress bar. Each file is

represented separately as there’s a separate request for each
one. This way, some small files will upload quickly, while
others load more slowly in parallel.
File list with progress bar for each one.

An Upload Form 333
<ul>
<li>
<span class="file-name">file.pdf</span>
<progress max="100" value="80">80% complete</progress>
</li>
<li>
<progress max="100" value="50">50% complete</progress>
</li>
</ul>
Supporting browsers display the <progress> element as

a progress bar. The element has two attributes: max and
value. The max attribute describes how much work there
is to be done. In our case, it’s set to 100 as we’re working in
percentages. The value specifies how much is complete,
which we’ll be updating with JavaScript.
$.ajax({
xhr: function() {
var xhr = new XMLHttpRequest();
xhr.upload.addEventListener('progress', function(e) {
if (e.lengthComputable) {
var percentComplete = e.loaded / e.total;
percentComplete = parseInt(percentComplete * 100);
li.find('progress')
.prop('value', percentComplete)
.text(percentComplete + '%');
}
}, false);
return xhr;
}
});
334 Chapter 8
As jQuery (at the time of writing) doesn’t expose the

onprogress event, we’ve created an XMLHttpRequest object
ourselves.
The handler first checks to see if the server has cor-

rectly sent a Content-Length header by seeing if
e.lengthComputable is true. If it is, then we can deter-
mine how much of the file has been uploaded, which is cal-
culated by dividing e.loaded by e.total. That value is then
converted to a percentage before updating the progress bar.
The progress bar’s inner text is also set. This is so users with
a browser that lacks support for the <progress> element
can still see it. That’s inclusive.
Success
Next, we want to show users when a file has been suc-

cessfully uploaded. First, the file name is converted into a
link so users can download and verify the file if they wish.
Second, we inject a success message of “File uploaded” and
a Remove button, which is useful if the file was uploaded
by mistake.
An Upload Form 335
File list with successfully uploaded files marked as such.
<li>
<a class="file-name" href="/path/to/file.pdf">file.pdf</
a>
<span class="success">
xlink:href="#tick"></use>
</svg>
File uploaded
</span>
<input type="submit" name="remove1" value="Remove">
</li>
$.ajax({
success: $.proxy(function(response){
if(response.file) {
li.html(this.getSuccessHtml(response.file));
}
}, this)
});
We’re using the success callback, which receives the

response from the server as an object. The response con-
tains a file property, which contains the path and name of
336 Chapter 8
the file. This is used to create the HTML that is injected into
the list item.
Note: The demo uses Multer3 and Express4 to process the

request and generate the response object. You can use
whatever you like.
Error
If the uploaded file is too big, or in the wrong format, we’ll

need to show users an error message. This is similar to the
success message, but instead of showing a green success
message with a tick, we’ll show a red message with a warn-
ing symbol. Note that the error markup below is the same
as the error markup used to show validation errors in a
standard form.
File list with unsuccessfully uploaded files marked with error messages.
3 http://smashed.by/multer
4 http://smashed.by/expressjs
An Upload Form 337
<li>
<span class="error">
xlink:href="#warning-icon"></use>
</svg>
File.pdf is too big.
</span>
<button type="button">Remove</button>
</li>
$.ajax({
success: $.proxy(function(response){
if(response.error) {
li.html(this.getErrorHtml(response.error));
} else if(response.file) {
li.html(this.getSuccessHtml(response.file));
}
}, this)
});
The updated success function now checks to see if the

response has an error. If it does, the error state will be con-
structed and injected into the list item instead.
Screen Reader Feedback
While the feedback is useful for sighted users, screen reader

users won’t hear any feedback. To provide a comparable
experience (inclusive design principle 1), we’ll need to use a
live region — something we’ve extensively covered in “A
Checkout Form” and “A Flight Booking Form.”
338 Chapter 8
<div class="visually-hidden" role="status" aria-

live="polite">
Uploading files. Please wait.
</div>
Note: The live region is visually hidden because sighted

users have already been catered for with the live progress bar
and its various states. The CSS for the visually-hidden
class is set out in “A Checkout Form” (on page 116).
The live region will change at various points:
When Description
Upload starts Uploading files. Please wait.
Upload completes [Name of file] has been uploaded.
Upload fails For example: [Name of file] is too big.
The size must be less than 2Mb.
FEATURE DETECTION AND INITIALIZATION

Feature detection was introduced in the very first chapter
and also demonstrated in “A Flight Booking Form” (on page
164). We’ll use it once again here because it’s important to
detect features before using them, otherwise we’ll create a
broken experience for users of browsers that lack support.
An Upload Form 339
The drag-and-drop enhancement uses a number of APIs that

not all browsers recognize. Here are the feature detection
functions with a usage example at the end.
function dragAndDropSupported() {
var div = document.createElement('div');
return typeof div.ondrop != 'undefined';
}
function formDataSupported() {
return typeof FormData == 'function';
}
function fileApiSupported() {
var input = document.createElement('input');
input.type = 'file';
return typeof input.files != "undefined";
};
if(dragAndDropSupported() && formDataSupported() &&
fileApiSupported()) {
var Dropzone = function(container) {
//...
};
}
There are three feature detection functions; one for each of

the features that browsers might not recognize. We then
make sure that there is support for all of them before defin-
ing our Dropzone component.
As the Dropzone is conditionally defined based on feature

detection, we need to detect the Dropzone function during
initialization too. If it’s defined, then the browser supports it,
340 Chapter 8
meaning it’s safe to initialize; otherwise users will get the

basic (but not broken!) experience.
if(typeof Dropzone != 'undefined') {

new Dropzone($('.dropzone'));
}
A NOTE ABOUT OLDER BROWSER SUPPORT

Uploading files immediately onchange and ondrop might
be confusing to users because, at least conventionally speak-
ing, forms are submitted with a separate action. However,
this isn’t only a conventional problem. It doesn’t work cross-
browser either.
For example, some older browsers won’t trigger the dialog

when the label is used as proxy,5 and while the onchange
event is supported there are two problems:
1. Choosing the same file (or a file with the same

name) for a second time, won’t fire the onchange
event, which creates a broken interface.6 The solu-
tion is to replace the entire file input after the event
with a clone of itself. As the cloned input would
need to be refocused, screen readers will announce
it for a second time, which is mildly annoying.
5 http://smashed.by/ieinput
6 http://smashed.by/selectevent
An Upload Form 341
2. The onchange event won’t fire until the field is

blurred.7 Newer browsers offer the oninput event,
which solves this problem because it fires the event
as soon as the value changes.
Whether you need to support such browsers depends on

your situation but it’s worth being aware of the problems.
Fortunately, the feature detection above happens to rule out
the offending browsers.
Other Considerations
There’s a number of additional design considerations for
uploading files, some of which are deep topics in their own
right. Let’s run through them quickly now for completeness.
CONVERT FORMATS AUTOMATICALLY

If users need to upload a spreadsheet, we should let them
upload proprietary formats, like Microsoft Excel, but also
non-proprietary ones like CSV.
Whichever they choose, we can convert it into the right

format when it’s processed. As you can see, the “Give
choice” principle is more than just offering different inter-
action modalities.
7 http://smashed.by/ieinput
342 Chapter 8
This is another example of doing the hard work so users

don’t have to.
MANY USERS STRUGGLE TO FIND FILES

Many users, particularly less digitally savvy users, may
struggle to use a file picker because they don’t know where
the file lives on their computer. And sometimes the file
needs to be transferred to the computer from an external
camera or device, creating additional effort and confusion.
Ed Horsford, a designer at GDS, conducted research around

this for the UK’s passport renewal service, which involves
users uploading a passport photo. He said that desktop
users who used a file importer (such as Windows photo
importer) struggled to find where the file was imported
when it came to selecting it from the file picker.
I’d class myself as digitally savvy, but I’ve also struggled to

locate files I’ve downloaded from different applications. My
personal workaround involves remembering to save my
files to desktop. This makes the files easy to find — I just
have to remember to delete the clutter every now and then.
Ed also said that while users generally found picking a

photo easy on mobile, some Android devices house the pho-
tos inside “Documents” and not a folder named “Photos,”
which research revealed threw most users.
An Upload Form 343
One simple way to help users with this is to provide addi-

tional guidance and instructions about how and where to
save files, or alternatively tell users where they’re likely to
find the file.
Upload form with a field hint explaining that “Your photo may be in your
Pictures, Photos, Downloads or Desktop folder. Or in an app like iPhoto.”
IT’S EASIER TO TAKE A PHOTO ON MOBILE

Your photos are taken and stored on your mobile device, so
it's easier to upload them using that device: there's no need
to transfer files from elsewhere. That's all well and good if
people are using your service on mobile, but what if they're
on desktop?
For services that require users to already have an account,

this is quite easy. If I create an advert on eBay but want to
344 Chapter 8
upload photos from my mobile, I just have to login on my

mobile to continue where I left off — my in-progress advert
will be there ready and waiting for me.
For services that don’t require being logged in, such as the
passport renewal service, it’s trickier. In such cases, consider
directing users to their phone with one-time security codes
or unique URLs that they can type easily into their phone.
Screen guiding users to continue the process on their mobile

with a special link.
THIRD-PARTY INTEGRATION
Some digitally savvy users may already use third-party
services, such as Dropbox, to store files. Giving users a way
to upload or provide files from these services may well be
easier for them, especially if your service is already con-
nected with theirs.
Be warned, however, that it may be unhelpful or even con-

fusing to users who don’t know what Dropbox is. Be sure
An Upload Form 345
to make the different choices clear, and test widely in user

research sessions.
MICROCOPY: “UPLOAD” OR “ATTACH”

Generally speaking, there are two ways to communicate to
users about uploading files. The first is to use “Attach,” but
this seems best suited for email. In almost every other situa-
tion “Upload” seems more common, which is what we used
for our generic drag-and-drop upload form earlier.
THE ACCEPT AND CAPTURE ATTRIBUTES

The file input has two interesting attributes that affect the
file uploading experience: accept and capture.
The accept Attribute
The accept attribute takes a string that indicates which

types of file the picker will accept.
<input type="file" accept="image/*">
When supported, the browser/device may offer users a more

stringent experience when choosing files. In Chrome and
Safari on iOS and Android, it will give users a choice of which
app to use to capture the image, including the option of taking
a photo with the camera or choosing an existing image file.
346 Chapter 8
iOS (left) and Android (right) dialogs for selecting images on a device when
the accept attribute is used.
But on desktop browsers it will prompt the user to upload

an image file from the file system, disabling files that aren’t
accepted (images in the above case). The problem is, users
won’t be told why the files are disabled as there’s no feedback.
The capture Attribute
The capture attribute, when supported, indicates the pref-

erence of getting an image from the camera:
<input type="file" accept="image/*" capture>

<input type="file" accept="image/*" capture="user">
<input type="file" accept="image/*" capture="environment">
An Upload Form 347
Adding the capture attribute without a value lets the

browser decide which camera to use (if there’s one avail-
able), while the user and environment values tell the
browser to prefer the front and rear cameras respectively.
The capture attribute works on Android and iOS, but is
ignored on desktop. Beware that on Android this means the
user will no longer have the option of choosing an existing
picture as the camera app will be started directly instead,
which is probably undesirable.
Summary
In this chapter we began by looking at the native file picker
as the browser gives us quite a bit of power for free. How-
ever, we also looked at the various problems that crop up
with multiple file uploads.
From there, we looked at various solutions that started

with the persistent upload form, before enhancing the
interface with a more ergonomic and inclusive drag-and-
drop interface.
348 Chapter 8
THINGS TO AVOID
• Prioritizing form over function.

• Using the multiple file input without considering
browsers that lack support for it.
• Replacing a standard file picker with a drag-and-
drop interface.
• Forcing users to use a particular file format, when
we can convert it for them automatically.
Demos
• Dropzone: http://smashed.by/dropzonedemo
350 Chapter 9
An Expense Form
As a self-employed freelancer I have to submit expenses
for my tax return. It’s a pain but if I do it correctly I get tax
breaks. The problem is that I have so many expenses to
enter and a limited amount of time to enter them.
The anatomy of an expense depends on the system you

use and, perhaps, the country you live in. It might include
a description, company name, date, amount, and proof of
purchase. How can we design a form that makes inputting
multiple entries easy, fast, performant, and inclusive?
Of course, if you know how many entries are needed
in advance, then give users a form with that number of
fields, make them required, and that’s about it. But if we
don’t know how many entries are needed in advance (like
expenses), keep reading.
The Persistent Form Pattern (Again)

In the previous chapter, I introduced the persistent form
pattern. In short, we gave users an upload form that users
can keep using until they’ve finished uploading as many
files as they need, at which point they can proceed or exit
the page, whichever is best (see page 324).
An Expense Form 351
Left: an upload form that stays on screen to be used as many times as

necessary. Right: file list above the upload form with uploaded files. Each file
can be deleted.
There are a number of other forms on the web that use the
persistent form pattern. For example, GitHub’s “Add collab-
orators” form and the infamous to-do list form that many
JavaScript frameworks use to demonstrate their approach.1
GitHub’s “Add collaborators” form.
This pattern works for adding expenses too. Each time the
user submits an expense, it will be added to the list that sits
above the form.
1 http://smashed.by/todomvc
352 Chapter 9
Left: expense form before adding an expense. Right: same form with a list of
added expenses above.
This pattern is well-suited to short, simple forms that can

be submitted in one go. The pattern does, however, have a
number of downsides:
1. Users might need to use a form that has dynamic ques-

tions (branching) that are conditionally shown based
on previous answers. In this case, the pattern doesn’t
work so well. We’ll look at branching in more detail
shortly.
2. As the list of added expenses grow, the form moves

further down the page. This could be a problem, espe-
cially on mobile, as users would have to scroll down to
see and use the form. Alternatively, you could put the
list below the form, but after adding an expense users
may not see the newly added item.
An Expense Form 353
3. Having multiple calls to action (to submit, and to

proceed or exit) might be confusing, especially for
cognitively impaired users. Where possible, one call to
action is preferable as it requires less thinking.
4. Each submission requires a separate request to the

server. This could be frustrating when lots of entries
need to be added.
Branching with One Thing Per Page

One of the major problems with the persistent form pattern
is that it can’t handle branching. Branching involves users
being asked different questions depending on previous
answers. For example, if users are expensing a car, they’ll
need to enter mileage; if they’re expensing a train ticket,
then they’ll need to enter its price.
In this case, the one thing per page pattern (first discussed
on page 70) is more suitable. This is because it presents one
question at time, meaning we can show users different
pages depending on their answers. This solves the branch-
ing problem elegantly and simply, but what if users need
to enter multiple expenses and submit them in one go?
We can add an additional screen to the end of the journey
asking users if they’d like to add another expense. Selecting
yes takes the user down the same flow again; selecting no
completes the task.
354 Chapter 9
The add expense flow using “one thing per page,” with a question at the end
of the flow asking if the user wants to add another one.
The Add Another Pattern

Both the persistent form pattern and the one thing per page
pattern suffer from the same problem — that each expense
created requires at least one trip to the server, which is slow.
How might we solve this problem?
The add another pattern works by giving users a single form,

on a single page, submitted in a single step. However, the
user can keep adding fields for as many expenses as they
need. For demonstration purposes, let’s simplify the anat-
omy of an expense down to just a description and cost.
HOW IT WORKS
The form starts with enough fields to enter one expense.
However, there’s an Add Another button, that when pressed,
will instantly clone the fields so that users can enter the
details of an additional expense.
An Expense Form 355
Left: add expense form with “Add another” button. Right: after having clicked
the button to clone the fields allowing the user to add multiple expenses.
Users can keep on doing this until they’re done, at which

point the user is able to submit all their expenses at once,
with just a single trip to the server — speeding up the pro-
cess drastically.
Note: The basic experience (before adding the JavaScript

enhancement), works the same way except that pressing
the “Add another” button will generate the new fields on
the server.
356 Chapter 9
THE BASIC MARKUP
<form>
<div class="addAnother">
<div class="addAnother-item">
<div class="field">
<label for="items[0][description]">
<span class="field-label">Description</span>
</label>
<input
type="text"
id="items[0][description]"
name="items[0][description]">
</div>
<div class="field">
<label for="items[0][amount]">
<span class="field-label">Amount</span>
</label>
<input
type="text"
id="items[0][amount]"
name="items[0][amount]">
</div>
</div>
<input type="submit" name="addAnother" value="Add
another expense">
</div>
<input type="submit" name="submitexpenses" value="Submit
expenses">
</form>
An Expense Form 357
There are two important notes about this form:
1. The expense fields are wrapped in a

<div class="addAnother-item">. This lets
us target the expense with CSS and JavaScript
(explained shortly).
2. The input’s name and id attributes have a special array-

like naming convention, which we’ll discuss now.
PROCESSING MULTIPLE (DYNAMIC) EXPENSES

When the form is submitted, the payload will consist of
multiple expenses. The server will need to process this
payload, but it won’t know how many expenses will be sent
ahead of time; that is, the amount of expenses is dynamic.
To help the server recognize and process the expenses, a con-

tract must be made between the client and the server. When
it comes to forms, the contract is forged by the name attribute.
<input type="text" name="items[0][description]">

<input type="text" name="items[0][amount]">
Note the special name attribute value. By formatting it this

way, the request payload can be parsed and converted into
an array of expenses that the server can process easily.
358 Chapter 9
Some server-side frameworks, such as Express,2 are

designed to recognize this convention. Let’s run through
each part and its meaning.
• items is the name of the group. You can use what-

ever name you like. On the server, this will be used
to identify the list of expenses.
• [0] represents the particular expense in the list and
starts from zero, like a JavaScript array. That is, the
second expense will be represented by [1], and so
on.
• [description] and [amount] represent the
specific attributes about the expense; in this case,
the description and the amount. For each unique
attribute you want to process, you’ll need a name to
identify it with.
ADDING ANOTHER EXPENSE

Pressing the “Add another” button needs to create a new set
of expense fields. There are several ways we might go about
doing this. For example, we might use templating — which
could be written in JavaScript3 or in HTML.4
2 http://smashed.by/expressjs
3 http://smashed.by/templateliterals
4 http://smashed.by/template
An Expense Form 359
Both approaches, however, lack browser support. A simple,

alternative approach involves cloning the already existing
expense fields.
function AddAnotherForm(container) {
container.on('click', '.addAnother-addButton',
$.proxy(this, 'onAddButtonClick'));
}
AddAnotherForm.prototype. {
var item = this.getNewItem();
this.getItems().last().after(item);
};
AddAnotherForm.prototype.getNewItem = function() {
return this.getItems().first().clone();
};
AddAnotherForm.prototype.getItems = function() {
return this.container.find('.addAnother-item');
};
There are three small functions that have been split out for
readability and maintainability. When the button is clicked,
we get a clone of the first <div class="addAnother-item>
in the form. Then we add the clone to the end of the form.
Injecting Remove Buttons
The form initially starts out with just a single expense.

There’s no “Remove” button because the user has to sub-
mit at least one expense. However, when the user adds
another expense, we need to add a remove button to the
360 Chapter 9
first expense. To do this, when the “Add another” button

is pressed, we’ll need to check whether a remove button
should be added.
// previous code
var firstItem = this.getItems().first();
if(!this.hasRemoveButton(firstItem)) {
this.createRemoveButton(firstItem);
}
};
AddAnotherForm.prototype.hasRemoveButton = function(item) {
return item.find('.addAnother-removeButton').length;
};
AddAnotherForm.prototype.createRemoveButton =
function(item) {
item.append('<button type="button" class=" addAnother-
removeButton">Remove</button>');
};
Now, when the button is clicked, the function checks to see if

the first expense has a remove button. If it doesn’t, one is cre-
ated. The reason we have to check for its existence is because
the first expense may or may not have a remove button,
depending on how many expenses the user has added.
Having given the first expense a remove button, we’ll need

to apply the same provision for the newly cloned expense,
like this:
An Expense Form 361
AddAnotherForm.prototype.getNewItem = function() {
var item = this.getItems().first().clone();
if(!this.hasRemoveButton(item)) {
this.createRemoveButton(item);
}
return item;
};
This function uses the same helper methods in the exact

same way. Now, whenever an item is cloned it will always be
cloned with a remove button.
Updating the Attributes
Having cloned the fields and ensured each field has a

remove button, we need to update the name and id attri-
butes. This ensures that the newly cloned fields adhere to
the naming convention so that the server can process the
submission (as explained earlier).
But how can our script know what name and id values to
use? To make this easy, we’ll store the naming convention
inside data attributes.
<input data-name="items[%index%][description]" data-

id="items[%index%][description]">
<input type="text" name="items[0][amount]">
362 Chapter 9
The reason for both the name and the id data attributes
is because some fields consist of multiple inputs with
the same name. For example, as laid out in chapter 2, “A
Checkout Form,” the name of each radio button is the same
because it identifies the set to which they belong. The id
identifies the individual radio button.
Now all we need to do is replace %index% with the new

index of the cloned item, like this:
AddAnotherForm.prototype.updateAttributes = function(index,
item) {
item.find('[data-name]').each(function(i, el) {
el.name = $(el).attr('data-name').replace(/%index%/,
index);
el.id = $(el).attr('data-id').replace(/%index%/,
index);
($(el).prev('label')[0] || $(el).parents('label')[0]).
htmlFor = el.id;
});
};
The function works by searching for all form controls that

have the data-name data attribute. For each control it finds,
it will update the control’s name and id attributes by replac-
ing %index% with the new index, which has to increase by 1
each time.
An Expense Form 363
Finally, the label’s for attribute is set to the control’s id

attribute. If we didn’t do this, then when the user clicks the
cloned label, focus will be moved to the first field instead.
Left: clicking the second description field sets focus to the first description
field erroneously.
Note: The code for retrieving the label uses a logic OR

operator. This is because the label appears in different
places depending on the type of field. In the case of a text
field, for example, it will be the previous sibling. However,
for radio buttons, should an expense need radio buttons, it
will be the parent.
Managing Focus
When the “Add another” button is clicked, it should focus the

first newly created form field. Screen readers will announce
the field, prompting the user to fill out the expense.
364 Chapter 9
// code
item.find('input, textarea, select').first().focus();
};
REMOVING AN EXPENSE
When the user clicks an item’s remove button, there’s a
number of tasks we need to implement. First, of course, is
that it should be removed from the form.
function AddAnotherForm(container) {
// code
this.container.on('click', '.addAnother-removeButton',
$.proxy(this, 'onRemoveButtonClick'));
}
AddAnotherForm.prototype. > {
$(e.currentTarget).parents('.addAnother-item').remove();
};
We’re using jQuery’s event delegation. This is useful

because we’re adding and removing the remove buttons
dynamically. Without using delegation we’d have to keep
adding and removing event listeners, which is a pain.
When the event listener is called, the remove button is

referenced by e.currentTarget. Then it’s just a matter
of searching for the parent container for the expense and
removing it.
An Expense Form 365
Removing the Remove Buttons
When the penultimate expense is removed, we need to

remove the first item’s remove button. The user shouldn’t be
able to remove every expense, because they must enter at
least one.
$(e.currentTarget).parents('.addAnother-item').remove();
var items = this.getItems();
if(items.length === 1) {
items.find('.addAnother-removeButton').remove();
}
};
The function checks to see if there’s just one expense item

left in the form; if there is, the remove button is removed.
Updating the Attributes
When the user has added a number of expenses, they are

free to remove any they might have provided by mistake. The
interface gives users control to remove any item they want.
Remember, the names of the fields use array-like indexes

so the form can be processed on the server. If there are
three expenses, for example, but the user deletes the second
expense, then the data will be out of sync.
366 Chapter 9
We can fix this by running through all the expenses in the

form and updating the fields’ indexes:
// code
var items = this.getItems();
// code
items.each($.proxy(function(index, el) {
this.updateAttributes(index, $(el));
}, this));
};
This loops through all the items in the form and invokes the
updateAttributes() method from earlier.
Managing Focus
When the user clicks the remove button, everything inside

the <div class="addAnother-item"> will be removed.
But what happens to the focus when you delete the cur-
rently focused element? Heydon Pickering answers this
question in “A Todo List”:5
[...] browsers don’t know where to place focus when it has been
“ destroyed in this way. Some maintain a sort of “ghost” focus where

the item used to exist, while others jump to focus the next focus-
5 http://smashed.by/a11ytodolist
An Expense Form 367
able element. Some flip out completely and default to focusing the
outer document — meaning keyboard users have to crawl [...] back
to where the removed element was.
We could set focus to the previous or next expense item, but

this seems arbitrary and confusing. Alternatively, we could
set focus to the “Add another” button, but that’s presumptu-
ous and odd — why delete an item if you’re just going to add
another one — users are better off just typing over the fields
that are already there.
Instead, we can set focus to the heading, which puts users

back to the beginning of the expense form while announc-
ing itself (“Expenses, heading, level 1,” or similar) to screen
reader users.
// code
this.container.find('.addAnother-heading').focus();
};
The problem is that, by default, headings aren’t focusable.

But we can fix that by giving the heading a tabindex="-1"
attribute. The -1 value allows us to focus the element pro-
grammatically using JavaScript, without making it user-fo-
cusable, using the Tab key, for example.
368 Chapter 9
<h1 tabindex="-1">Expenses</h1>
When the heading is focused, browsers will give the head-

ing an outline. This should be removed because the head-
ing is not an interactive element and so shouldn’t appear
interactive.
.addAnother-heading { outline: none; }
Once the heading is focused, pressing Tab will focus the

first form field, which makes orientation simple.
FEEDBACK
As it stands, the act of adding and removing expenses
provides sufficient feedback for sighted users: items in
the form can be seen to appear or disappear from the list.
Giving users an additional notification bar would draw
users’ attention in two directions. And as more expenses
are added, the notification bar will be out of the viewport so
users wouldn’t see it anyway.
Screen reader users are also catered for because the act
of adding and removing items moves the users focus and
announces the focused element accordingly. This may not
tell the user explicitly that the item has been added, but it
should be enough. If research shows otherwise, you can add
An Expense Form 369
a hidden live region (as set out in chapter 3, “A Flight Book-

ing Form”) to give screen reader users explicit feedback.
Animation Isn’t Necessarily Valuable
It’s possible that judicious animation effects can help users

understand an interface.
But all too often, designers want to add animation for the sake
of it. Users just want to get things done. Needless animation
is jarring and actually detracts from the user experience. Like
anything else, animation should only be used if it adds value.
Even when animation is valuable to some users, it can be

harmful to users with cognitive impairments, such as atten-
tion deficit hyperactivity disorder (ADHD) and autism.6
In the case of the expense form, had we not moved focus

from the button to the field, an animation might mean the
item’s arrival is more likely to be noticed. Alas, we are moving
focus to the form field and so animation is unnecessary here.
6 http://smashed.by/needlessanimations
370 Chapter 9
Summary
In this chapter we looked at three different patterns that
let users submit multiple expenses into a system. Really
though, expenses were used just for demonstration pur-
poses — these patterns are applicable to all kinds of data, not
just expenses.
The persistent form and one thing per page patterns are more
suitable for infrequent use and users with a lower digital
literacy. The add another pattern is more suitable for frequent
usage that doesn’t require branching.
The downside to the add another pattern is that despite the

usability provisions we’ve put in place, the interface is a
little more complicated to operate.
There’s no right or wrong here. It’s about choosing the most

appropriate pattern for your users.
An Expense Form 371
THINGS TO AVOID
• Letting browsers manage the focus when the focus

element is removed from the document.
• Giving non-interactive (but programmatically fo-
cusable elements) an outline when focused.
• Giving users multiple sources of feedback for a
single action.
• Animating parts of the interface to no practical
purpose.
Demos
• Add another: http://smashed.by/todolistdemo
372 Chapter 10
A Really Long and

Complicated Form
D
ifferent types of tasks take different amounts of
time to complete. The one thing per page pattern
(introduced in chapter 2, “A Checkout Form”) helps
users complete tasks in one sitting, but what about those
that take hours, or even days, to complete?
In MailChimp, for example, I’ll usually start drafting an

email campaign weeks before I send it. There are a number
of steps to complete, and in a particular order too. First, I
check the content reads well. Then I need to make sure it
looks good in various email clients. Then days or weeks
later, I’ll run through some final checks, decide the subject
line and schedule it for release.
Other tasks might be performed by several people using

the same application. For example, processing a return may
involve someone receiving the goods at the warehouse;
then a decision maker may look at the goods to make sure it
satisfies the returns policy.
Some services, such as applying for a mortgage or register-

ing for a bank account, involve supplying personal informa-
tion, and this may involve offline processes such as sending
identification documents in the post.
A Really Long and Complicated Form 373
How can we design forms that play nicely with this complex
and long-form process that can take weeks to complete? And
by different people across digital and non-digital journeys?
First, though, it must be said that if you can simplify the

process so it’s short — or even remove it altogether — do
that. If not, this chapter explores some useful patterns spe-
cifically designed to solve these things.
The Check Before You Start Pattern

One of the best ways we can help users save time is by not
wasting it in the first place. One way to do that is to tell
users what they need to know before they start the process.
For example, to apply for an HSBC mortgage in the UK, users

must: consent to a credit check; confirm the property is in a
habitable condition; and be aged 18 or over. HSBC tells users
this on the first page of the application, before they start.
This is helpful, but users may also want to know:

• what the overall process involves
• if they need to visit a branch
• how long it takes to get a decision
• what documents they’ll need (and what alternative
options they might have, if any)
374 Chapter 10
The first page of the HSBC mortgage application process explaining what
users need to know before applying.
Knowing this up front not only saves users a lot of time,

but it can also lower the operating costs of the service: by
reducing the time and effort support teams spend working
out and explaining all of this to people over the phone.
Other processes and services are more complicated and can

depend on the individual’s circumstances. For example, the
Renew Your Passport service (shown below) needs to ask
a series of questions to determine someone’s eligibility to
apply for a passport.1
1 http://smashed.by/renewpassport
Renew Your Passport service asking questions to determine the user’s best
course of action. The final screen tells users what to do having told the service
that their password has been stolen.
If the applicant has lost their passport, for example, then they
aren’t able to use the main service. The Government Digital
Service calls this “Check Before You Start,” which you can
read about in “We’ve published the check before you start
pattern” by Harry Trimble and Rob Le Quesne.2
The Task List Pattern

In “The Psychology of Checklists,” Lauren Marchese explains
the importance of breaking down big tasks into smaller ones,
which is proven to motivate people.3 When we experience
even small amounts of success, our brains release a chemical
called dopamine, which gives us feelings of pleasure, learn-
ing, and motivation.
2 http://smashed.by/govukchecklist
3 http://smashed.by/smallgoals
376 Chapter 10
Most of us work in teams employing agile methodologies.

This involves breaking down a large project into epics,
stories, and tasks. Complete enough tasks and the story is
done. Complete enough stories and the epic is done. Com-
plete enough epics and the project is done. Of course, our
work is never done, but you know what I mean.
What’s really happening is that tasks seem far easier to

achieve when they’re broken down. Crucially, if tasks are
small enough, then we’ll get that hit of dopamine more fre-
quently, which creates momentum. Momentum improves
morale, and morale improves productivity.
That’s not all checklists are good for. In The Design of Everyday
Things, Don Norman says:
Checklists are powerful tools, proven to increase the accuracy of
“ behavior and to reduce error [...]. They are especially important in

situations with multiple, complex requirements, and even more so
where there are interruptions.
As people perform digital tasks on the go using different

devices, the chances of interruption are high. Designing for
interruption and being able to jump back into the middle
of a long and complex task is crucial. As mentioned earlier,
some parts of the process happen offline too.
The task list pattern, as coined by the Government Digital

Service (GDS), shows a page with several top-level tasks.4
Each top-level task is broken down into several subtasks.
Each one of those takes users through a flow — whether
each flow consists of one or several screens doesn’t really
matter, as long as it’s achievable in a reasonable time. Once
a subtask is completed, users come back to the task list view
with that particular task marked as complete.
The GDS task list pattern with all tasks marked as completed.
This is not a pattern reserved for government services:

MailChimp users have a similar need. The only difference is
the visual design and the flatter hierarchy of tasks.
Instead of text, MailChimp uses iconography to mark tasks

as complete. I discussed the pros and cons of iconography in
chapter 3, “A Flight Booking Form” (see page 184).
4 http://smashed.by/govuktasklist
378 Chapter 10
Additionally, instead of standard-looking links, they use call-

to-action buttons labeled as “Resolve” to prompt the user
and reduce the effort on their part.
MailChimp’s campaign task list page showing that two tasks still need to be
completed
The exact design details you choose will come down to your
product’s design language and user research, but it’s key to
ensure that:
• each task status is clearly marked so users can see

what’s left at a glance
• users can get a feel for how long is left until com-
pletion
• previous information is saved, so that users can
return easily to it later
WHEN ALL TASKS ARE COMPLETE

When the last task in the list is completed, the user can still
go back to the task list page with all the tasks marked as com-
pleted. This let’s users bask in the glory of completing every-
thing (which is another hit of dopamine). After all, the best
part of completing a long to-do list is seeing all those ticks.
It also gives users a chance to review all their answers and

make any amendments if necessary, which speaks to inclu-
sive design principle 4, “Give control.” The only other thing
to remember is to give users an obvious way to proceed
once everything is complete. Offer users a clear, single call
to action.
ADDITIONAL CONSIDERATIONS
The points discussed above are probably applicable to any
very long form you’re designing, but you might also want
to consider:
• Explaining what users need, such as documents, in

the context of each individual task.
• Indicating how long each task will take. An esti-
mate or a range can work well. If you can’t offer this
information, then you may need to break down the
tasks further.
380 Chapter 10
• Using verbs for task names. For example, “Agree to

the terms,” “Create subject line,” “Choose template.”
• Listing tasks in order. If so, use an ordered list, the
advantages of which are discussed in chapter 5, “An
Inbox.”
• Marking who needs to complete the task. This is
only useful if the tasks are performed by different
people.
• Sizing all the tasks the same. Don’t take this too
literally, but if one task is twenty questions and
another is two, then take another look.
Summary
In this chapter we looked at two important patterns. The
check before you start pattern makes a process transparent
and saves users a lot of time. It can even be designed for a
range of personal circumstances by asking users a series of
questions about themselves.
We then looked at how the task list pattern breaks down a

very long process into a series of smaller, more manageable
tasks that users can return to in their own time.
These patterns can make a very long and very complex jour-
ney relatively easy to complete.
CHECKLIST
• Avoid really long and complex forms if you can.

• Tell users as much as they need to know up front
so they can decide whether they are eligible for a
service.
• In more complex situations, ask users a series of
questions to determine the best course of action,
and save their time and the operating costs of the
service.
• Break up large tasks into smaller pieces, and allow
users to save their progress so they can return later
easily.
Thank you for reading my book!
One of the best things about this topic is that there are
always going to be new challenges to solve and new pat-
terns to define. If you can think of a new form problem
you’d like me to look at, please send me a message on
Twitter (@adambsilver). You never know — with enough new
problems, a sequel could be on the cards.
382 Index
Disabling/hiding buttons 63, 237

Index Drag-and-Drop 340
Dragover and dragleave 329
Dropdown menus 124
3rd-party integration 345 Email Address Field 33
Accept attribute 346 Error messages 53, 57, 63
Accessibility 20, 68 Expense Form 351
Accordions (checkout) 68 Feature Detection 164, 339
Adaptive design 241 File upload 315
Address field 79, 82, 105 Filters 150, 280,
Ajax 296, 302 303, 311
Analytics (search) 265 Flight Booking Form 123
Auto-Capitalization 203 Float Labels 25
Auto-Tabbing 206 Focus styles 33, 253, 364
Autocomplete (countries) 127 Forgotten password link 213
Autocorrect 203 Form Feedback (errors) 51
Autofill (credit card) 97 Form Field Width 81
Back button / links 119, 298 Form Labels 20, 31
Before You Start Pattern 374 Form Placeholders 22
Billing/shipping address 105 Form Submission 50, 58, 240,
Button Text (Call to Action) 48 290
Character countdown 90 Form Validation 49, 59
Checkboxes 193, 226 Guest Checkout 74
Checklist affirmation pattern 62 Hiding labels 269
Checkout Form 68 Highlighting marked emails 231
Choosing a Flight 186 Hover versus click 246
Choosing A Seat 190 Html5 validation 50
Collapsible Filters 303 IE problems 46, 86
Checkout Confirmation 109 Inclusive Design Principles xv
Confirming versus undoing 259 Inline erros 56
Country input/autocomplete 127, 152 Inline validation 61
CVC input (credit cards) 99, 104 Input mistakes 60, 104
Date picker 155, 158 Interactive filters 280
Delivery notes 88 JavaScript 240
Delivery Options 84 Keyboard interaction 135, 146, 174,
Denoting Selected Filters 311 206, 253
Design Principles xv Label position 31
Index 383
Legend element 85 Remove Buttons 360, 366

Limiting text 89 Responsive Design 241, 308
List Types 221 Responsive Menu 238
Live regions 92, 134 Returns policy 265
Loading states 296 Review Page (checkout) 107
Login Form 200 Screen reader enhancements 52, 92, 116,
matchMedia API 250 134, 269
Mobile Phone field input 77 Search Form 263
Multiple File Picker 321 Search input 127
Multiple submit buttons 233 Search Results 275
No password sign-in 28 Second-Time Experience 112
Number input 77, 97 Select All 254
Offer choice 83 Select box 124, 132, 239
Onchange event 240 Show password 40
One Thing Per Page 212, 354 Social Login 215
Optional fields 78 Stepper component 181
Optional or required fields 79 Sticky menus 235
Order Summary (checkout) 117 Submit Button 208, 233, 268
Ordered and unordered lists 225 Success Messages 256
Password Field Design 204 Tab order 137
Password reveal 39 Tables 223
Payment forms (credit cards) 94 Task List Pattern 376
Persistent Form Pattern 324, 351 Toast messages 257
Phone number input 77 Toggles 40, 105, 271
Photo upload 344 Undo pattern 260
Placeholder 22 Upload Form 315
Postcode field 81 Username Label 201
Privacy 216 Username or Password 209
progress bar / Indicator 113 Doesn’t Match
Progressive enhancement 35, 52, 82, Users May Not Notice the 300
(code examples) 130, 164, Results Update
183, 273, Validation 187
305, 328, visually-hidden 116, 135, 229,
340 269
Radio buttons 84, 87, 126, When to Fly 154
193 Where to Fly 123
Registration Form 18 ZIP code field 81
More From Smashing Magazine
• Apps For All: Coding Accessible Web Applications

by Heydon Pickering
• Art Direction For The Web (Oct. 2018)

by Andy Clarke
• Design Systems
by Alla Kholmatova
• Digital Adaptation
by Paul Boag
• Inclusive Design Patterns

by Heydon Pickering
• Smashing Book #6: New Frontiers in Web Design

Written by Laura Elizabeth, Marcy Sutton, Rachel
Andrew, Mike Riethmueller, Harry Roberts, and others.
• The Sketch Handbook

by Christian Krammer
• User Experience Revolution

by Paul Boag
Visit smashingmagazine.com/printed-books/ for our full

list of titles.

Form Design Patterns

Uploaded by

Copyright:

Available Formats

Form Design Patterns

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Form Design Patterns

Uploaded by

Copyright:

Available Formats

In memory of my

beautiful and amazing

Cover design: Espen Brunborg

Form Design Patterns was written by Adam Silver

Please send errors to: errata@smashingmagazine.com

3 A Flight Booking Form 123

4 A Login Form 200

6 A Search Form 262

7 A Filter Form 278

8 An Upload Form 314

9 An Expense Form 350

10 A Really Long and Complicated Form 372

He’s particularly interested in inclusive design and design

About The Reviewer

Heydon has written multiple books on the subject of

Graham Veal for helping to set up the foundations of the

Finally, my wife Jen, for her patience in letting me spend so

Making a habit of flouting convention will garner you

Paradoxically, in a world saturated with rule breaking and

Let me give you an example. Imagine my excitement when

That’s actually radical, and really refreshing to read. Because

Don’t get me wrong: I’m not saying there’s nothing new in

To me, this book is about simple solutions to would-be

My obsession with forms started when — like with any

Forms allow users to create, update and delete things.

This is a big part of why I’m writing this book. Typically,

First, instead of solving the same problem from scratch

Second, by solving the same problem in the same way,

Using design patterns for digital experiences or, more

Why These Forms?

There are a few problems with this approach to design.

Instead of centering the book on principles, I decided to

Here’s a chapter rundown.

We’ll start with a basic registration form and take a look

The one thing per page design pattern is a cornerstone of cre-

3. A FLIGHT BOOKING FORM

We’ll dive into the world of progressively enhanced, custom

trols at length, and look at breaking away from convention

We’ll look at the ubiquitous login form. Despite its simple

We’ll design ways to manage and action email in bulk, our

We’ll create a responsive search form that is readily availa-

Users often need to filter a large set of unwieldy search

give up. Filters pose a number of interesting and unique

Many services, like photo sharing, messaging, and many

We’ll investigate the special problem of needing to create

10. A REALLY LONG AND COMPLICATED FORM

What About Principles?

But where should our principles come from? We can either

Our principles normally stem from a belief system. We

This book is about designing forms for the web. It would be

Frank Chimero talks about this at length in “The Web’s

Whatever we build, in the end, is about users. I don’t want

These principles are about good design — and good design

1. Provide a comparable experience. Ensure your inter-

2. Consider situation. People use your interface in

3. Be consistent. Use familiar conventions and apply

4. Give control. Ensure people are in control. People

5. Offer choice. Consider providing different ways for

6. Prioritize content. Help users focus on core tasks,