Signifiers

Also called Signifier, Design signifier

Quick answer

A signifier is a perceivable cue that shows people which action is possible and where to do it. A flat plate on a door shows where to push, and a raised edge on a button shows where to press. Don Norman proposed the term in 2008 to separate what a thing can do from the cue that shows it.

~20 sec read

011 min

Why signifiers matter: the door that needs a sign

Picture a person carrying a tray who walks up to a glass door. Each side of the door has the same vertical handle. The door opens one way only, but nothing on it says which way. The person pulls, nothing moves, and they try again. Then they notice a small printed sign that says push.

Designers call a door like this a Norman door, after Don Norman, who wrote his book The Design of Everyday Things partly because of problems like it. A Norman door gives people the wrong signals so strongly that it needs a printed sign to say whether to push or pull.

The door itself works. It opens when pushed, and anyone who already knows how it works can use it. What is missing is a visible cue at the spot where the hand lands. Norman's word for such a cue is signifier. He needed a new word because the old one, affordance, had started to mean two different things.

Two printed stickers side by side on a glass door, one with the word PULL and one with the word PUSH, in light letters on black.
Fig. 1A glass door with two printed signs, one reading pull and one reading push.Robert S. Donovan from Adams, NY, USA · CC BY 2.0 · Wikimedia Commons
doorpullsignpush
Follow the arrow from the hand on the handle to the printed sign. The handle gave no signal, so the sign had to.

022 min

Where the word signifier came from

The old word, affordance, came from a psychologist who was not thinking about doors. J. J. Gibson used affordance for an action that a person or animal can do in a given place. In Gibson's sense, an affordance exists whether or not anyone can see it or knows about it.

Norman brought the word into design in his 1988 book. Looking back in 2008, he wrote that Gibson's affordances did not have to be perceivable, but the affordances he himself had meant in 1988 were the ones people can perceive. He added that the word had been used loosely ever since, and that this had caused confusion. He also admitted that part of the blame was his, because he should have written perceived affordance.

Many designers said they had added an affordance when they put a button or an icon on a screen. Norman called this a misuse. On a screen, a person can click anywhere, so a button does not add the possibility of clicking. What the button gives is a target, a convention people have learned, and feedback.

In 2008 Norman wrote a short column for the magazine Interactions with a blunt title, "Signifiers, Not Affordances". He defined a signifier as any physically perceivable cue, whether it is put there on purpose or arrives by accident. His advice to designers fitted in one line: stop talking about affordances and provide signifiers.

032 min

How signifiers work on a screen

The door shows the idea with a physical object, but most signifiers today sit on screens. Norman wrote that on a screen the designer controls only perceived affordances, because the computer already lets a person point at and click any spot. So almost every cue on a screen is something people have learned from other screens.

Take a link. A hand-shaped pointer over it, blue text and an underline are all learned cues, and none of them changes what a click can do. Norman called the pointer change a convention, because the person can still click anywhere whatever the pointer looks like. A raised button works the same way. Its shadow does not make the button pressable. It tells a person who has seen raised buttons before that this one is.

Cues come in a few common forms:

  • A label says the outcome in words. A button that reads "Pay now" tells people what happens after the press.
  • An icon stands for an action people already know, such as a magnifying glass for search.
  • Shape and contrast set a control apart from static text. A filled button on a plain background reads as pressable, and a greyed button reads as unavailable.
  • Position puts the cue where the action happens. A plate sits at hand height on a door, and a label sits on or beside its control.
  • Sound and touch carry the cue when the eyes are busy, such as a short tone when a payment goes through or a buzz in a phone.

Together these cues narrow the gulf of execution, the gap between what a person wants to do and the actions the system offers.

A cue and a function can disagree in two ways. The function can exist with no cue, as with a swipe that deletes a message and shows no sign beforehand. Or the cue can exist with no function, as with a handle on a door that only opens by pushing. Both leave the person guessing. Because most cues are learned, they also depend on habit and culture. A colour or a gesture can mean one thing in one country and another elsewhere, so a cue that works for one group needs checking with the next.

042 min

Signifiers nobody designed

Not every signifier is placed by a designer. Norman's 2008 column opens with cues that nobody made for anyone. Say you run up the stairs of a station to catch a train and reach a platform with no train on it. Did you miss it, or has it not come yet? The state of the platform tells you. People waiting mean the train has not arrived, and an empty platform means it has gone. Norman calls this an incidental signifier, a cue that shows something as a side effect of what other people are doing.

When an incidental signifier misleads

Such a cue is useful, but it is not fully reliable. Norman notes that it works better in a small town with few trains than in a crowded city, where many trains use the same platforms. A designer who relies on an accident accepts that the cue will sometimes be wrong.

Desire lines: trails across a field

A second case is the desire line. When people's footsteps wear a trail across a field, city planners read the trail as a sign that people want a paved path there. Many steps in the same line wear a trail, and the trail shows the next walker the shortcut. Nobody planned it, yet it works as a signifier.

Wear marks in documents

Researchers have copied the idea into software. Norman cites the usage marks called edit wear and read wear, proposed in 1992 by Hill, Hollan, Wroblewski and McCandless, as another way to add signifiers that guide readers of electronic documents.

A narrow strip of bare soil runs through thick green grass beside a red brick wall, with a street and parked cars on the right.
Fig. 2Footsteps alone wore this trail. Nobody planned it, yet it shows the next walker where to go.Punker1999 · CC0 · Wikimedia Commons
stepstrailpath
Read left to right: steps, then a trail, then the path that does not exist yet. The trail is the cue that points to it.

052 min

What weak signifiers cost: the flat design study

Designers can also lose signifiers by accident, and the cost has been measured. Flat design is a style that removes depth effects such as shadows and gradients from interface parts. It replaced skeuomorphic design, which draws controls to look like physical objects. Many flat pages ended up with buttons that look like plain text or like empty outlines.

In 2017 Kate Moran and her colleagues at Nielsen Norman Group tested what that does. They took nine pages from live websites and made two nearly identical versions of each. One version had strong signifiers, such as underlined blue text or a slightly raised button. The other had weak ones, such as an outlined button with no fill or link text styled like plain text. A total of 71 people each saw one version of every page and said "I found it" when they saw where they would click.

Pages with weak signifiers took 22% more time and drew 25% more fixations than the same pages with strong ones. A fixation is a moment when the eyes rest on one spot, so more of them means more looking around. People did see the weak controls. What changed is that they were less sure of them and kept searching.

When flat design still works

The cost was not the same on every page. It shrank on pages with little content, a traditional layout and targets that stood out from their surroundings. A sparse page with a high-contrast button in the usual place can use a flat style with little loss. A crowded page or an unfamiliar layout cannot.

Finding is not discovering

The tasks named one control to find, so the study measures findability. That differs from discoverability, which is whether people notice a feature they were not looking for. Nielsen Norman Group notes that strong signifiers matter most when the goal is for people to find out about a feature they did not know existed. A weak cue costs a person who is looking for the control a little time. It can cost a person who is not looking the whole feature.

061 min

When a signifier is missing: hidden navigation

The same loss of signifiers shows up when a control is hidden altogether. In 2016 Kara Pernice and Raluca Budiu of Nielsen Norman Group ran tests with 179 people on six websites, on phones and on desktops. Some sites showed their main navigation links in plain view. Others put them behind a menu icon, often the hamburger icon made of three stacked lines.

When the navigation was hidden, content discoverability dropped by more than 20%. Here discoverability means whether people could find content through the navigation when the home page did not link to it. On desktop, people were also at least 39% slower.

A hamburger icon is a signifier, but it signifies only that a menu exists. It shows nothing about what the menu holds, so the links inside give no cue until someone opens it. Sites with visible links, or a mix of visible and hidden links, did better in the test.

071 min

Applying signifiers in a design

The studies suggest a short list of moves for adding signifiers.

  • Make a clickable element differ from the text around it in more than one way, such as colour plus an underline or a raised edge. One weak cue is easy to miss.
  • Put the cue where the action happens. On the door in the opening case, a flat plate at hand height does the same job as the printed sign, and a label on or beside a control does it on a screen.
  • Reuse what people already know. People bring habits from other products, which Jakob's Law describes, so a link that looks and behaves like links elsewhere needs no teaching.
  • Show state. A control that cannot be used right now should look different from one that can.
  • Test by asking, not by guessing. Show a page and ask people to say where they would click, the same task Nielsen Norman Group used.

Signifiers can also be overdone. If everything on a page looks pressable, people cannot tell the main action from the rest, and the page loses its visual hierarchy. A cue can also promise something false, such as a shape that looks pressable but does nothing. A person who clicks it learns that the page's cues cannot be trusted. Keep the number of strong cues small, and make each one true.

081 min

Accessibility and signifiers

A signifier that depends on seeing colour fails for some people. Two success criteria in the Web Content Accessibility Guidelines (WCAG) 2.2 cover this. Success criterion 1.4.1, Use of Color, says colour cannot be the only visual way to show an action or tell one element from another. Success criterion 1.4.11, Non-text Contrast, asks for a contrast ratio of at least 3:1 between the parts that identify a control and the colours next to them, except for controls that are inactive.

These rules protect people with colour blindness and people with low vision. A control marked only by a red or green label gives them no cue at all. A pale grey outline on a white page is the kind of case the second criterion is meant to catch. Add a second cue, such as an underline, a border or an icon, and check the contrast of that cue.

092 min

Signifiers vs nearby concepts

Signifiers get mixed up with a few related ideas, and the main one is affordance. The door in the opening case can be pushed, which is its affordance. A flat plate where the hand lands would be its signifier. One is about what is possible and the other is about what shows it. A signifier can even exist without the affordance, as with a handle on a door that only opens by pushing, which is why the two need separate names.

ConceptWhat it isThe question it answers
This pageSignifiersWhere do I act, and how?What it is: The perceivable cue that shows an action
DesignAffordanceCan this be done at all?What it is: The action that is possible
DesignDiscoverabilityWould a person ever find this?What it is: How easily people notice that a feature exists
DesignNatural MappingWhich control goes with which thing?What it is: A control layout that matches the layout of the thing it controls

Natural mapping answers a different question: which control belongs to which thing. A good mapping can also serve as a signifier, because the layout itself shows the link.

The other meaning of the word signifier

Search for the word and many results are about linguistics. In the theory of the linguist Ferdinand de Saussure, the written word open on a shop door is a signifier, and the idea that the shop is accepting customers is its signified. That sense is older and separate from the design one. In design, a signifier is a cue that shows an action, not a word that stands for an idea.

?4 questions

Questions people ask

Is a signifier the same as an icon?

No. An icon is one kind of visual cue. A signifier can be any perceivable cue, including a label, a shape, a position, a sound, a texture or a worn trail.

Do signifiers only matter on screens?

No. Norman's own examples are a train platform, a bookmark and a trail across a field. Screens make them matter more, because nothing physical shows what a control does until the designer adds a cue.

Can a signifier be wrong?

Yes. A cue that points to an action the thing does not support makes people act wrongly, like a handle on a door that only opens by pushing. The cue and the real function must agree.

Does flat design remove signifiers?

Only when it removes the cues that mark a control. A flat button can still stand out through colour, shape and position, and the Nielsen Norman Group study found smaller costs on sparse pages with clear targets.

§9 sources

Sources

Show all 9 sources
  1. Don Norman, "Signifiers, not affordances", Interactions, November–December 2008. jnd.org

  2. Don Norman, "Affordance, Conventions and Design (Part 2)", first published in Interactions, May–June 1999. jnd.org

  3. Don Norman, "Affordances and Design", 17 November 2008. jnd.org

  4. Kate Moran, "Flat UI Elements Attract Less Attention and Cause Uncertainty", Nielsen Norman Group, 3 September 2017. nngroup.com

  5. Kara Pernice and Raluca Budiu, "Hamburger Menus and Hidden Navigation Hurt UX Metrics", Nielsen Norman Group, 26 June 2016. nngroup.com

  6. W3C, "Understanding Success Criterion 1.4.11: Non-text Contrast", WCAG 2.2. w3.org

  7. W3C, "Understanding Success Criterion 1.4.1: Use of Color", WCAG 2.2. w3.org

  8. Daniel Chandler, "Semiotics for Beginners: Signs". aber.ac.uk

  9. 99% Invisible, "Norman Doors: Don't Know Whether to Push or Pull? Blame Design." 99percentinvisible.org

Keep reading

More from Design

All of Design
All of Design

Missing a concept?

Tell us what to write next. We’ll email you when it’s live.

Several? Separate them with commas.

Only to tell you when it’s live.

How we use your email: privacy.