| name | usability-test-plan |
| description | Plan a usability test — define research questions, tasks, participant criteria, and analysis approach. Produces a structured test plan ready for execution. Use before conducting moderated or unmoderated usability testing with real users. |
| argument-hint | [feature or flow to test] |
| user-invocable | true |
| allowed-tools | Read, Write, Edit, Bash, Glob, Grep |
Usability Test Plan
Plan a usability test for $ARGUMENTS. This skill produces a structured test plan covering research questions, methodology, participants, tasks, and analysis. Use this before running any moderated or unmoderated usability test — a test without a plan produces anecdotes, not insights.
Step 1: Define Research Questions
Start with what you want to learn. Research questions drive every subsequent decision — methodology, tasks, participants, and analysis.
Define 3-5 specific, answerable research questions:
### Research Questions
| # | Question | Type | What it will tell us |
|---|---|---|---|
| RQ1 | [Specific question — e.g., "Can users complete checkout without assistance?"] | Behavioural / Attitudinal | [Decision this informs] |
| RQ2 | [e.g., "Where do users get confused in the onboarding flow?"] | Behavioural | [Decision this informs] |
| RQ3 | [e.g., "Do users understand what the pricing tiers include?"] | Comprehension | [Decision this informs] |
| RQ4 | [e.g., "How does the new navigation compare to the current one?"] | Comparative | [Decision this informs] |
Good research questions are:
- Specific — "Can users find the export button?" not "Is the UI good?"
- Answerable — observable through user behaviour, not speculation
- Decision-linked — the answer changes what we build
Output: 3-5 research questions with types and the decisions they inform.
Step 2: Choose Methodology
Select the testing approach based on research questions and constraints:
### Methodology
| Dimension | Choice | Rationale |
|---|---|---|
| **Moderation** | Moderated / Unmoderated | [Why — moderated for exploration, unmoderated for scale] |
| **Location** | Remote / In-person | [Why — remote for reach, in-person for context] |
| **Protocol** | Think-aloud / Task-completion / A/B comparison | [Why — think-aloud for discovery, task-completion for benchmarking] |
| **Prototype fidelity** | Live product / High-fidelity prototype / Low-fidelity wireframe | [Why — match to development stage] |
| **Tool** | [UserTesting / Lookback / Maze / in-person with recording] | [Why — capabilities needed] |
Method selection guide:
- Exploration (understanding problems) → Moderated think-aloud
- Validation (confirming solutions) → Unmoderated task-completion
- Benchmarking (measuring performance) → Unmoderated with quantitative metrics
- Comparison (A vs B) → Within-subjects or between-subjects design
Output: Methodology table with rationale for each choice.
Step 3: Define Participant Criteria
Determine who to test with and how to find them:
### Participants
| Criterion | Requirement |
|---|---|
| **Number** | [5-8 for qualitative; 20+ for quantitative benchmarking] |
| **User type** | [Match to personas — e.g., "Small business owner, 1-10 employees"] |
| **Experience level** | [Novice / Intermediate / Expert with the product or domain] |
| **Demographics** | [Relevant demographics — age range, accessibility needs, tech literacy] |
| **Exclusions** | [Who to exclude — employees, recent participants, competitors] |
### Screener Questions
| # | Question | Accept | Reject |
|---|---|---|---|
| S1 | [Screening question — e.g., "How often do you use project management tools?"] | [Daily/Weekly] | [Never] |
| S2 | [e.g., "What is your role?"] | [Manager, Team Lead] | [Developer, Designer] |
| S3 | [e.g., "Have you used [product] before?"] | [Yes, in last 30 days] | [Never used it] |
### Recruitment
| Item | Detail |
|---|---|
| **Source** | [Customer panel / Recruitment agency / Social media / In-app intercept] |
| **Incentive** | [Amount and form — e.g., "$50 gift card"] |
| **Timeline** | [Recruitment start → sessions complete] |
Nielsen's research shows 5 participants catch approximately 85% of usability issues. Use 5 for qualitative discovery, more for quantitative benchmarking.
Output: Participant criteria, screener questions, and recruitment plan.
Step 4: Design Tasks
Create realistic scenarios that map to research questions:
### Task Design
| # | Task | Scenario | Success criteria | Time limit | Research question |
|---|---|---|---|---|---|
| T1 | [Action-oriented task name] | [Realistic context — "You want to invite a colleague to your project..."] | [Observable outcome — "User reaches the invitation confirmation screen"] | [minutes] | RQ1 |
| T2 | ... | ... | ... | ... | RQ2 |
| T3 | ... | ... | ... | ... | RQ1, RQ3 |
Task design rules:
- Scenario, not instruction. "You need to change your billing address" not "Click Settings, then Billing, then Edit Address"
- One goal per task. Compound tasks confuse analysis — split them
- Order matters. Start with simple tasks to build confidence, increase difficulty gradually
- Include a deliberately difficult task. One task should push the interface — this reveals breaking points
- Cover all research questions. Every RQ maps to at least one task; every task maps to at least one RQ
Output: Task table with scenarios, success criteria, time limits, and RQ mapping.
Step 5: Write Moderator Guide
Script the session to ensure consistency across participants:
### Moderator Guide
#### Introduction (5 minutes)
- Welcome and thank participant
- Explain purpose: "We're testing the [product/feature], not you — there are no wrong answers"
- Confirm recording consent
- Explain think-aloud protocol (if applicable): "Please say what you're thinking as you work through the tasks"
- Ask: "Do you have any questions before we start?"
#### Warm-up (3 minutes)
- [Background question — "Tell me about how you currently [relevant activity]"]
- [Familiarity question — "How often do you [relevant task]?"]
#### Tasks (30-40 minutes)
For each task:
1. Read the scenario aloud (or share written scenario for unmoderated)
2. Observe without intervening
3. Note: time to complete, errors, hesitations, verbal comments
4. If stuck for > [time limit]: offer one neutral prompt ("What are you looking for?")
**Follow-up probes (use after each task):**
- "What did you expect to happen there?"
- "Was anything confusing or surprising?"
- "How would you rate the difficulty of that task? (1-5)"
#### Debrief (5-10 minutes)
- "Which task was most difficult? Why?"
- "What would you change about this experience?"
- "Is there anything else you'd like to share?"
- Thank participant, provide incentive
Output: Complete moderator guide with introduction, tasks, probes, and debrief.
Step 6: Define Metrics and Analysis Plan
Decide what to measure and how to analyse it before running sessions:
### Quantitative Metrics
| Metric | Definition | Target | How measured |
|---|---|---|---|
| **Task success rate** | % of participants completing the task | > 80% | Binary: completed / not completed |
| **Time on task** | Seconds from task start to completion | < [target]s | Stopwatch / tool timer |
| **Error rate** | Number of wrong actions per task | < 2 per task | Observer count |
| **Satisfaction (SEQ)** | [Single Ease Question](https://measuringu.com/seq10/) — "How easy was this task?" (1-7) | > 5.5 | Post-task questionnaire |
| **System satisfaction (SUS)** | [System Usability Scale](https://measuringu.com/sus/) — post-test | > 68 (above average) | Post-test questionnaire |
### Qualitative Analysis
| Method | Description |
|---|---|
| **Affinity diagramming** | Group observations into themes across participants |
| **Severity rating** | Rate each issue: [Critical / Major / Minor / Cosmetic] |
| **Frequency count** | How many participants encountered each issue |
| **Rainbow spreadsheet** | Task × Participant matrix showing success/failure/assistance patterns |
### Severity Scale
| Level | Definition | Action |
|---|---|---|
| **Critical** | User cannot complete the task | Must fix before release |
| **Major** | User completes with significant difficulty or errors | Should fix before release |
| **Minor** | User notices but works around it | Fix in next iteration |
| **Cosmetic** | Noticed only when pointed out | Fix if time permits |
Output: Metrics table with targets, qualitative analysis approach, and severity scale.
Step 7: Plan Logistics
### Logistics
| Item | Detail |
|---|---|
| **Tool** | [Recording/testing platform] |
| **Recording** | [Screen + audio / Screen + audio + video / Notes only] |
| **Consent form** | [Template — covers recording, data use, withdrawal rights] |
| **Session duration** | [minutes — typically 45-60 for moderated, 15-30 for unmoderated] |
| **Schedule** | [Date range, sessions per day — max 4 moderated sessions/day to avoid fatigue] |
| **Observers** | [Who watches — product, design, engineering; max 2 observers per session] |
| **Note-taking** | [Dedicated note-taker or recording-only] |
| **Pilot session** | [Date — run one pilot to test the guide before real sessions] |
| **Deliverable** | [Report format and delivery date] |
Always run a pilot session with a colleague or friendly user before real sessions. The pilot tests your test — unclear tasks, timing issues, and technical problems.
Output: Logistics checklist with dates, tools, and responsibilities.
Rules
- Research questions drive everything. If you cannot connect a task to a research question, remove it. If you cannot connect a research question to a decision, remove it.
- 5 participants catches 85% of usability issues. Do not delay testing to recruit 20 people for qualitative research. Test early with 5, iterate, test again.
- Tasks must be realistic scenarios, not instructions. "Find how to export your data" is a scenario. "Click the gear icon and select Export" is a walkthrough. Walkthroughs test memory, not usability.
- Never lead the participant. "Do you think this button is hard to find?" is leading. "Tell me what you're looking for" is neutral. The moderator guide must contain only neutral probes.
- Define analysis before running sessions. Deciding how to analyse data after collection introduces bias. The metrics and severity scale are decided in the plan, not the report.
- Pilot the test before running it. An untested test plan wastes participants. Always run one pilot session.
Output Format
# Usability Test Plan: [Feature/Flow Name]
**Date:** [date] | **Researcher:** [name] | **Status:** [Draft/Approved/In progress/Complete]
## 1. Research Questions
[From Step 1 — 3-5 questions with types]
## 2. Methodology
[From Step 2 — moderation, location, protocol, tools]
## 3. Participants
[From Step 3 — criteria, screener, recruitment]
## 4. Tasks
[From Step 4 — scenarios with success criteria]
## 5. Moderator Guide
[From Step 5 — introduction, tasks, probes, debrief]
## 6. Metrics & Analysis
[From Step 6 — quantitative targets, qualitative approach, severity scale]
## 7. Logistics
[From Step 7 — schedule, tools, pilot, deliverables]
Related Skills
/ux-researcher:usability-review — heuristic evaluation without users. Use as a complement: heuristic review finds obvious issues cheaply, usability testing finds issues only real users reveal./ux-researcher:persona-definition — participant criteria should align with defined personas. Define personas first if they don't exist.
