Usability Test

---
name: usability-test
description: Plan and run a usability test around realistic user tasks, representative participants, observable success, moderator discipline, evidence capture, and prioritized product implications.
---
# Usability Test

Use when a product flow, prototype, or implementation needs evidence about whether intended users can understand and complete important tasks.

## Procedure
1. Define the product decisions, target participants, critical tasks, prototype or build revision, and what successful task completion means.
2. Create realistic scenarios that give participants a goal without telling them which control or path to use.
3. Prepare starting state, test data, environment, recording or note method, consent, and any moderator handoff rules.
4. Ask participants to work naturally and avoid rescuing them until the point of intervention is itself recorded as evidence.
5. Observe task success, route, hesitation, errors, recovery, terminology confusion, missed information, repeated attempts, and requests for help.
6. Use follow-up questions after behavior occurs to understand intent and interpretation without leading the action beforehand.
7. Distinguish interface defects from prototype limitations, participant-specific unfamiliarity, and technical failures unrelated to the design.
8. Synthesize patterns and severe outliers by task impact and confidence, then hand product implications to the appropriate design or product owner.

## Decision rules
- A test task should describe the user's goal, not the UI steps.
- Moderator assistance changes the evidence and must be recorded.
- Five participants is not a universal rule; sample needs depend on variation, method, and claim.
- Usability testing finds interaction evidence, not population-level preference percentages from a small qualitative sample.

## Quality gate
The test is complete when participant and build context are recorded, realistic tasks produce observable behavior, moderator influence is visible, findings distinguish design from test-environment artifacts, and prioritized implications trace back to specific evidence.