cognitive psychology
Norman's seven stages of action
What are Norman's seven stages of action?
Norman's seven stages of action break a single interaction into the steps your mind actually goes through: form a goal, then plan, specify and perform (the doing), then perceive, interpret and compare (the checking). They map the two gulfs - execution and evaluation - into a cycle, giving designers a checklist of seven distinct places where an interface can help or fail.
Also known as: seven stages of action, norman's stages of action, action cycle, stages of action
Prefer to watch?
Watch the recap 0:51
The demo
One simple action - turning the music up - is secretly seven steps. Walk through them and watch a “trivial” button-press unfold into the whole cycle of doing and checking.
What this demo shows (text version)
A single action - "turn the music up" - is stepped through Norman's seven stages. Execution (the doing): form the goal (it's too quiet, I want it louder), plan the intention (I'll turn it up), specify the action (find and use the volume control), perform it (drag the slider). Evaluation (the checking): perceive the new state (the number/bar changed), interpret it (it's at 70 now), compare to the goal (yes, that's loud enough). The stages are grouped by the two gulfs they cross.
The point is diagnostic: a single press is really seven mental steps, and a design can fail at any one - no obvious control (can't specify), no feedback (can't perceive), ambiguous feedback (can't interpret). Locating which stage breaks turns "this is confusing" into a precise fix. The first four stages cross the gulf of execution; the last three cross the gulf of evaluation.
One action is really seven micro-steps in a loop: goal → plan → specify → perform → perceive → interpret → compare. The first four cross the gulf of execution (working out and doing the action); the last three cross the gulf of evaluation (perceiving and making sense of the result). Each is a place a design can break - so when something feels hard, find which stage is failing, and you've found what to fix.
Walk one tiny action - "turn the music up" - through all seven stages and you can feel how much invisible work a single button-press really is: a goal, a plan, picking the control, doing it, noticing the change, reading it, judging whether it matched. Each step is a chance for the design to help or trip you - which is exactly why "just press the button" so often isn't simple at all.
The seven stages run: form the goal (what you want), plan an intention (decide to act), specify the action sequence (work out exactly what to do), perform it (do it) - that's execution; then perceive the state of the world (notice what changed), interpret it (make sense of it), and compare it to the goal (did it work?) - that's evaluation. Then you loop, with the comparison feeding the next goal.
The value is diagnostic. When an interaction feels hard, the model lets you locate exactly where: couldn't form a goal (unclear what's possible), couldn't specify the action (no obvious control), couldn't perceive the result (no feedback), couldn't interpret it (ambiguous feedback). Each failing stage points to a different fix, so the seven stages turn "this is confusing" into a precise problem statement.
It's the engine behind the gulfs of execution and evaluation - the first four stages cross one, the last three cross the other - and underpins much of usability heuristics. You don't run the checklist for every screen, but keeping it in mind catches the stages designers most often forget: making the next action discoverable, and making the result of an action unmistakable.