Friction as a design tool
Effort suppresses behaviour, which makes it a tool for slowing down the wrong actions.
Friction is any cost a flow imposes: steps, fields, waits, decisions, confirmations. Adding friction reduces how many people complete an action, which is a problem when the action is good for them and a feature when it is not. The craft is applying it where a pause changes the outcome and removing it everywhere else.
Completion falls as steps accumulate
How it shows up in software
Friction appears as confirmation modals, typed-name deletion, undo windows, cooling-off periods and rate limits. It appears unintentionally as a login wall, a required field nobody reads, or a third page of a checkout. Most products have friction in the wrong place: heavy on the safe path, light on the destructive one.
Using it well
- Reserve friction for actions that are irreversible or expensive to undo: deleting an account, sending money, publishing to everyone.
- Prefer undo to confirm. A ten second undo window on a sent email costs nobody anything and catches the real mistakes.
- Make the friction specific to the risk. Typing the repository name to delete it works because it cannot be clicked through on autopilot.
- Count the steps on your core path once a quarter and delete one. Friction accumulates without anyone deciding to add it.
Where it turns manipulative
- Friction placed on the exit rather than the risk is sludge: Friction added deliberately to stop someone doing a thing they are entitled to do.. Ten steps to cancel and one to subscribe tells you who the friction serves.
- Dressing up an obstacle as protection, such as forcing a call to a retention agent in the name of account security, is a lie about purpose.
- Overusing confirmations trains users to click through everything, which removes the protection from the one dialog that mattered.
Where you have seen it
GitHub repository deletion
Deleting requires typing the full repository name and reauthenticating, which makes the action impossible to complete inattentively.
Gmail Undo Send
A short window after sending where one click retracts the message, replacing a pre-send confirmation with a post-send reversal.
Coinbase and other exchanges' address whitelisting
New withdrawal addresses sit in a hold period before large transfers are allowed, so a compromised session cannot drain funds instantly.
What the research says
- Bettinger, Long, Oreopoulos and Sanbonmatsu, 2012 (Quarterly Journal of Economics)Well evidenced
Families who got help prefilling the US student aid form were substantially more likely to submit it and to enrol in college. Information alone, without the form help, did almost nothing.
The cleanest demonstration that the paperwork was the barrier, not the knowledge. It is evidence about removing friction, and it is the strongest single result in this entry.
- Cox et al., 2016 (CHI)Mixed evidence
Framed design friction as a deliberate interface technique and reviewed cases where a small interruption shifts users from automatic to reflective behaviour.
Mostly conceptual with small studies. It gives useful vocabulary; it does not tell you how much friction is right.
- Rogers et al. and the confirmation dialog literatureWell evidenced
Repeated confirmations get clicked through. Users learn to dismiss a dialog they see often, so the pause stops producing a pause.
Habituation is the main reason friction decays. A speed bump on a daily road becomes invisible within weeks.
Grades are a judgement about the evidence, not about how useful the idea is. Plenty of contested effects are still worth knowing, as long as you do not cite them as settled.