Meaningful Friction

Also called Intentional Friction, Productive Friction, Positive Friction

Quick answer

Meaningful friction is an effort, delay or extra step that a designer adds on purpose before an action that matters. It slows the person down so that they switch from fast, automatic thinking to deliberate thinking and choose with attention. It belongs where a mistake is costly or hard to undo, and it should not block routine tasks.

~20 sec read

011 min

Why Meaningful Friction goes against usability advice

Picture a designer at a file-storage app. A product manager shows them a funnel report: many people who start deleting a shared folder leave at the confirmation step. The manager asks the designer to remove that step.

The request follows normal usability advice. That advice treats every extra step as a cost, because friction makes people give up on tasks they wanted to finish. For most of an interface, the advice is right.

The funnel report is missing something, though. It counts the people who left, but it does not say why. Some of them may have left because the step was annoying. Others may have stopped because the step made them notice that they had picked the wrong folder, and for those people the step worked.

That second group is what meaningful friction is about. Some steps cost the product users, and some steps protect users from their own mistakes. A drop-off number cannot tell the two apart.

021 min

Where Meaningful Friction comes from

The research version of the idea has a date. In 2016, Anna Cox, Sandy Gould, Marta Cecchinato, Ioanna Iacovides and Ian Renfree, then at University College London, published a short paper at the CHI conference, the main yearly meeting on human-computer interaction. Its title was Design Frictions for Mindful Interactions.

The paper starts from the common view of the time. Design writing said that points of difficulty in using technology, called design frictions, should be removed so that people do not get frustrated and leave. The authors disagreed for many situations. They argued that a small obstacle built in on purpose can interrupt thoughtless, automatic use and prompt a moment of reflection. They listed benefits that run from fewer data-entry errors to support for changing health habits.

The phrase meaningful friction comes from practitioner writing, and no single person is known to have coined it. So this entry describes a design principle with research behind it, not a law with a formula.

032 min

Two meanings of Meaningful Friction

Writers use the word meaningful in two ways, and mixing them causes confusion.

In the first meaning, friction is meaningful because it protects a decision. A pause comes before an action that is costly or hard to undo, and the person gets a chance to check what they are about to do. This is the meaning Cox and colleagues argued for, and it is the main subject of this entry.

In the second meaning, friction is meaningful because effort makes the result feel worth more. A person who assembles a shelf, learns a hard level in a game or grinds coffee by hand may value the result more because of the work. Cox and colleagues note that games depend on designed challenge, and they separate it from a breakdown caused by poor interface design. Commentary under the name friction-maxxing makes a wider argument, that people should choose less convenient options in daily life on purpose. The agency RAPP gave product examples in 2023, such as two-factor authentication for security and a travel site that offers little sorting, so people browse more listings. Those examples are claims about what a team intended, and the article offers no measurements for them.

The two meanings need different evidence. Protective friction can be tested by counting the mistakes it prevents. Friction that adds value is harder to test, and a later section shows a condition under which it fails.

042 min

How Meaningful Friction works

The protective meaning rests on a model with two modes of thinking, which Cox and colleagues take from cognitive psychology. The first mode is fast and automatic, and the article on System 1 thinking describes it. It handles most of what people do on a phone without much attention. The second mode is slow and deliberate, and people use it when a situation needs focus.

An interface that makes every action one tap keeps people in the first mode. In that mode, two kinds of error become likely. People answer quickly and make mistakes, and they respond to cues such as an email notification in ways that do not match what they value, such as checking work email on a weekend.

The authors propose a remedy called a microboundary. It is a small obstacle placed before an action, which stops a person rushing from one context into the next. It costs a short amount of time. In that time the person can notice what they are about to do and move to the second mode. A microboundary does not forbid the action. The person can still go ahead.

Microboundary versus dark pattern

Cox and colleagues contrast their idea with dark patterns, which are interface designs that steer people into behaviour they may not want. A dark pattern keeps a person in the automatic mode so that they act as the company wants, while a microboundary pushes them out of it so that they can choose for themselves. The difference is whose interest the pause serves.

A short lockout gives time to plan

One form of microboundary is the task lockout, where the interface blocks a step for a short time. Studies that Cox and colleagues summarise found that lockouts improve attention. The pause stops people from clicking through message boxes without reading them, and it gives them time to plan their next action before they carry it out.

A less clear font raised accuracy

A second experiment, by Soboczenski and colleagues and described in the same paper, asked people to copy numbers from a source into an interface. When the source was shown in a less clear font, participants copied the numbers more accurately, and the task took no longer in any way the researchers could measure. The experiment's authors explained this by saying that the harder text pushed people toward a slower and more careful way of copying.

051 min

How Meaningful Friction shows up in real products

The designer can look at products that already make this choice. They show that a pause can sit in different places.

GitHub asks for effort at one moment only. To delete a repository, the owner opens the settings, reads the warnings and types the repository's name into a text box. Typing the name cannot be done by reflex, because the person has to look at the name and compare it with the one they meant. The cost is paid only on a rare action.

Friction and a restore window work together

GitHub's page also says that some deleted repositories can be restored within 90 days. That is a second design choice. A pause before the action and a way to reverse it afterwards cover different mistakes. The pause catches the mistakes a person would notice with a second look, such as the wrong repository. The restore window covers the mistakes nobody could notice in advance.

Gmail moved the pause to after the action. In 2009 the Gmail team added Undo Send as an experiment. Gmail holds a sent message for five seconds, and during that time the person can click Undo and return to the draft. Nothing is blocked, and if the browser closes during the wait, Gmail still sends the message. The person pays no cost unless they regret sending.

held5secondssent
Look at the right window: it is dashed because nothing has been sent yet. The message waits in the left window while the person can still change their mind.

062 min

What measured results show about Meaningful Friction

Those examples show designs, not results. Measured results are rarer, and three of them are worth reading.

A prompt before an offensive reply

Twitter, as it was then called, tested a prompt for people about to post a reply that matched its definition of offensive content. The prompt showed the draft and offered three choices: edit it, delete it or send it as written. In a 2022 report, Matthew Katsaros, Kathy Yang and Lauren Fratamico found that people who saw the prompt posted 6% fewer offensive tweets than people who did not. Of the prompted replies, 69% were sent as written, 9% were cancelled and 22% were revised. Most people sent anyway, so the prompt worked by changing a minority of decisions.

Browser warnings and clicks

The same pattern appears in security. In 2013, Devdatta Akhawe and Adrienne Porter Felt studied browser warnings using data from Firefox and Chrome. Chrome's SSL warning was bypassed 70.2% of the time, against 33.0% for Firefox's. Chrome needed one click to bypass the warning and Firefox needed three, but the authors did not think the extra clicks explained the gap. Their conclusion was that the design of a warning can change what people do, so a bigger obstacle is not automatically a better one.

A tool that works because it is effortful

Cox and colleagues also look at food diaries. Logging a meal soon after eating has been linked to weight loss, and so has simply taking a photograph of the food. They argue that the effort is the active part, because it forces a moment of deciding whether to eat what is on the plate. They warn that wearable cameras that log meals automatically would remove that step.

prompted repliessent69%cancelled9%revised22%
Compare the three bars: they share one scale, and most prompted replies were still sent as written.

072 min

Applying Meaningful Friction

The designer now has to decide about the delete step. Three questions sort the steps that deserve a pause from the ones that do not.

  • Is the action hard to undo? A pause is worth its cost when a mistake costs a lot, such as deleting shared files or sending money.
  • Is the action likely to happen without much thought? Actions that people repeat many times a day are the likeliest to be done by habit.
  • Would the pause give the person something to decide with? A pause that only repeats "Are you sure?" adds time and no information.

If the answers are yes, these moves help:

  • Put the pause just before the point of no return, and nowhere else.
  • Ask the person to look at something, not only to click. Typing the name of the folder works because the person has to read the name. A second confirm button does not.
  • Prefer an undo or a restore window when the mistake can be reversed. Some consumer laws use the same idea over days, because in some countries a buyer can cancel a purchase after making it.
  • Say why the pause is there, in one line. A pause with no stated reason looks like a bug.
  • Measure what the pause prevents, such as mistaken deletions and requests to restore files, next to the drop-off it causes. Either number alone is misleading.
  • Check who pays. The same step costs more for someone who uses a screen reader or only a keyboard.

For the file-storage designer, the answers point to a split. Deleting a shared folder gets a typed-name step. Deleting a single file gets an undo, which costs nothing unless it is needed.

082 min

When Meaningful Friction fails

A pause helps only while it still makes people think, and four things stop it from doing that.

Repeated prompts lose their effect

In the Twitter study, people were most likely to edit or drop a tweet at the first prompt. With each later prompt they were more likely to post it unchanged. A pause that appears on every action turns into one more step to click through. Let frequent users switch the pause off for low-risk cases, and keep it for rare, costly ones.

Long waits send people elsewhere

Lockouts work only if people use the wait to think about the next step. Cox and colleagues cite work showing that when a lockout is too long, people spend the wait on other work, and the pause no longer prompts them to think about the action. Keep the delay as short as the decision needs.

Effort only adds value when the task is completed

Norton, Mochon and Ariely tested the second meaning of the word in 2012. In four studies, participants assembled IKEA boxes, folded origami and built Lego sets, and they valued their own work more because of it. This is the IKEA effect. The effect appeared only when participants completed the task; it faded when they built something and then saw it destroyed, or could not finish it. Friction that adds value has to lead to a finished result, or it only frustrates.

When the pause is sludge

Cass Sunstein uses the word sludge for friction that is out of proportion or has no good reason, such as paperwork that costs people time or money. He recommends audits, in which an organisation lists its frictions, counts what each one costs and decides which to reduce. The test for meaningful friction is the reverse of sludge: the person can say what the pause is for, and it prevents something they would regret. A cancellation flow that is much harder than the sign-up flow fails that test, because the pause protects the company's revenue and not the person's decision.

091 min

Meaningful Friction vs. nearby concepts

Four neighbours are easy to mix up with meaningful friction. One question separates them: who benefits from the obstacle?

ConceptWhat it isWho it serves
Meaningful frictionA deliberate pause before an action that mattersThe person
Friction in generalAny effort, delay or difficulty, deliberate or notNobody in particular
SludgeFriction with no good reason, or out of proportionUsually the organisation
Dark patternsDesigns that steer people into behaviour they may not wantThe company

A forcing function is a fifth neighbour. It removes the choice: an interlock that stops a machine from starting while its guard is open leaves the operator no way to go on. Meaningful friction leaves the choice with the person.

?3 questions

Questions people ask

What should I say when a stakeholder asks to remove a pause?

Ask what the pause prevents. If nobody can name a mistake it stops, it may be sludge and can go. If someone can, keep it and measure that mistake.

Is meaningful friction the same as making a product harder to use?

No. A product can add a pause at one costly decision and stay easy everywhere else. The aim is one well-placed pause, not general effort.

Should frequent users see the same pause every time?

Not for low-risk actions. A repeated prompt loses its effect, so let frequent users switch it off where mistakes are cheap, and keep it for rare and costly actions.

§9 sources

Sources

Show all 9 sources
  1. Anna L. Cox, Sandy J. J. Gould, Marta E. Cecchinato, Ioanna Iacovides and Ian Renfree, "Design Frictions for Mindful Interactions: The Case for Microboundaries," CHI EA 2016.

  2. Matthew Katsaros, Kathy Yang and Lauren Fratamico, "Reconsidering Tweets: Intervening during Tweet Creation Decreases Offensive Content," ICWSM 2022.

  3. Devdatta Akhawe and Adrienne Porter Felt, "Alice in Warningland: A Large-Scale Field Study of Browser Security Warning Effectiveness," USENIX Security 2013.

  4. Michael I. Norton, Daniel Mochon and Dan Ariely, "The IKEA Effect: When Labor Leads to Love," Journal of Consumer Psychology, 2012 (open copy).

  5. Cass R. Sunstein, "Sludge Audits," Behavioural Public Policy, 2022.

  6. GitHub Docs, "Deleting a repository."

  7. Michael Leggett, "New in Labs: Undo Send," Official Gmail Blog, 2009.

  8. Clyde Zhang, "Examining meaningful friction in UX design," RAPP, 2023.

  9. Wikipedia, "Friction-maxxing" (a summary of recent commentary).

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.