Cognitive walkthrough checks whether a new user can infer each next action
Aliases: cognitive walkthrough · action inference · novice task analysis
What it is
A cognitive walkthrough follows the action sequence for a new user's task and asks whether the user will form the right subgoal, notice an available action, associate it with the goal, and interpret feedback as progress. It evaluates inference during early learning rather than inviting holistic expert opinion.
Why it happens
New users lack product-specific routines and rely on task knowledge, labels, layout, and feedback. The chain breaks when a goal is not activated, a control is hidden, mapping is ambiguous, or feedback is nondiagnostic. Stepwise questions localize the inference failure behind “hard to use.”
Studying it
Fix the user background, starting state, success state, and action paths. Have evaluators answer independently with evidence before synthesis. Compare predictions with novice testing to estimate misses and false positives; evaluator agreement does not replace user evidence.
Where it stops holding
Expert efficiency, long-term strategy, aesthetics, and emotion are not primary outcomes. A single canonical sequence creates false problems when multiple paths are reasonable. Experts familiar with the system can underestimate missing novice concepts.
Applying it
- Specify a novice profile, start, goal, and all defensible paths.
- At each step inspect goal, visibility, mapping, and feedback with screen evidence.
- Convert breakdowns into testable label, salience, sequence, or feedback hypotheses.
- Retest severe predictions with target novices performing independently.