interaction design
Friction
Is friction always bad in UX?
Friction is any effort an interface puts between a person and what they're trying to do - a step, a confirmation, a wait, a decision. It isn't simply bad: friction on the wrong action repels people, but a deliberate dose on a risky one is a safety net. The skill is matching the friction to the stakes, not removing it everywhere.
Also known as: ux friction, good friction, friction in design
Prefer to watch?
Watch the recap 0:45
The demo
Two actions, one routine and one destructive. Decide which deserves a confirm step - then read the verdict, and notice that the same friction helps one and hurts the other.
What this demo shows (text version)
Two actions sit side by side: a routine, easily-undone one ("mark all as read") and a destructive, irreversible one ("delete the project"). Each has a toggle for whether doing it should require a confirmation step, and a button to perform it - which, when friction is on, asks you to confirm first.
A verdict updates as you set them. The best configuration leaves the routine action frictionless (so it's fast) and puts friction on the destructive one (so it can't happen by accident); the worst does the opposite - slowing the routine task while leaving the dangerous one a single slip away. The lesson: friction isn't inherently good or bad, it's well-placed or misplaced - match it to the stakes and reversibility of each action.
Friction isn't good or bad in itself - it's misplaced or well-placed. Strip it from routine, low-stakes actions so the happy path flies; add a measured dose to destructive or irreversible ones so people can't sleepwalk into harm. Judge every bit of friction by the stakes of the action it guards, not by a blanket "less friction is always better".
Frictionless, the routine action flew - and the destructive one was a single slip from disaster. Add a confirm step and the destructive action became safe, but the routine one turned into a needless chore. Same friction, opposite effect: it's not about how much, but where.
Most of the time, less friction wins - every needless step on a routine, reversible action costs speed and patience and quietly sheds people who didn't need slowing down. That's the friction worth hunting and removing: the kind that guards nothing.
But some friction earns its keep. A "type DELETE to confirm", a pause before sending money, a review step before a permanent change - these are deliberate speed bumps that turn a careless slip into a considered choice. Remove them and you haven't made the product nicer; you've made its worst mistakes easier.
Match friction to stakes and reversibility. High-stakes or irreversible: add friction proportional to the harm. Low-stakes and easily undone: strip it to nothing and let an undo carry the rare mistake. The wrong move is treating all friction as waste - or sprinkling confirmations on everything until people click through them blind.