| name | Usability Quick Research |
| description | Child skill under Build Quality UI/UX for decision-first lightweight UX research. Sizes the bet before picking a method (bet-size → method matrix), enforces hypothesis-not-validation discipline, runs Krug-style 3-user moderated tests with a leading-question lint, audits old research and participant quality, and returns a research plan with stop conditions instead of a report. |
| skill_type | strategy |
| altitude | domain |
| activation | progressive |
| parent_id | build_quality_ui_ux |
| depth | 1 |
| preserve_markdown | true |
| legacy_paths | ["product-and-design.usability-evaluation-and-quick-research.skill","product-and-design/usability-evaluation-and-quick-research","docs/research/youtube-library/skill-drafts/usability-evaluation-and-quick-research/SKILL.md"] |
| path | apps/web/src/lib/services/agentic-chat/tools/skills/definitions/usability_quick_research/SKILL.md |
Usability Quick Research
Identity
Use this child when a UI decision needs user evidence instead of another internal opinion. Operating thesis (Hall): research is the act of becoming informed enough to make a better decision — start with goals and decisions, never with methods. Never answer "what research method should we use?" until the decision and unknowns are named.
This is a strategy skill at domain altitude: its dominant job is to decide how big the bet is and which is the smallest credible method, sizing risk before any method is picked; the ordered Workflow and the fill-in test script are the secondary procedure that serves that decision.
Activation
- Planning quick research, interview questions, usability tests, prototype tests, assumption checks, or lightweight synthesis.
- The user asks how to validate a screen, reduce UX risk, learn from users, or run a practical usability pass without a large research program.
- A design review depends on unknown user goals, language, confidence, or task success.
- Stakeholders are resisting research, leaning on old research, or asking for validation after a decision is already made.
Judgment
Bet Size → Method Selection
"The amount of research depends on how large the decision is, how much is already known, and how uncertain the team is" — Hall states this verbatim (BayCHI primary source), so the bet-size grouping is now a primary rule, not a derivation. Method rows are drawn from the Hall/Krug method catalog; Krug's reversibility limit on small-sample testing sets the High-tier floor.
| Bet size | Decision profile | Smallest credible methods | Floor |
|---|
| Low | Reversible, cheap to be wrong, ships this week | Desk/topical research, old-research audit, trunk test, heuristic pass, support-log or analytics review, competitive teardown | Self-serve, same day; no recruiting required |
| Medium | Sprint-level; reversible but costly to redo | 3-user moderated think-aloud test (script below), 1–2 semi-structured user interviews, prototype/Figma walkthrough with 2 users, card sort for IA questions | Hypothesis + disconfirming-evidence list written before contact; real or explicitly-justified proxy participants |
| High | Expensive to reverse: pricing, payment flows, security UX, repositioning, platform choice | All of the above, repeated rounds, plus quantitative companions (task success rate, time-on-task, error rate) on larger samples; stakeholder research first |
