ux-research（用户体验研究）

UX Research

Plan, conduct, and analyze user research that produces actionable insights for product design. This skill provides frameworks, methods, and practical guidance for every stage of the research process — from defining the right question to presenting findings that drive decisions.

When to Use This Skill

Use this skill when:

• Planning a research study (choosing methods, defining scope, writing a plan)
• Writing interview scripts or discussion guides
• Creating usability test plans with tasks and success metrics
• Recruiting participants and writing screener surveys
• Analyzing qualitative data (affinity diagrams, thematic analysis)
• Synthesizing research into personas, journey maps, or opportunity maps
• Setting up continuous discovery practices (weekly interviews, OSTs)
• Presenting findings and recommendations to stakeholders

Core Principle

Research is a tool for reducing risk, not a deliverable. The goal is always to inform decisions — not to produce documents. Research is not asking people what they like, not a political tool, and not science. It's applied inquiry that helps you solve the right problem.

Research Types

Choose your research type based on what decisions are in play:

Generative / Exploratory

When: Before you know what you're building. You need to define the problem. Methods: Interviews, field observation, contextual inquiry, diary studies, literature review Output: Problem definition, opportunity areas, personas, mental models

Descriptive

When: You have a design problem and need to fully understand the context. Methods: Contextual inquiry, fly-on-the-wall observation, journey mapping, card sorting Output: Current-state maps, workflow diagrams, mental models

Evaluative

When: You have potential solutions and need to test them. Methods: Usability testing, heuristic evaluation, A/B testing, click testing, tree testing Output: Severity-ranked issues, task completion rates, recommendations

Causal

When: A solution is live and you need to understand actual usage patterns. Methods: Analytics review, A/B testing, funnel analysis, surveys Output: Conversion data, behavioral patterns, quantitative validation

The Research Process

1. Define the Problem

Start with a clear problem statement. Base it on a verb that indicates outcome:

• Good: "Identify the barriers preventing users from completing onboarding"
• Good: "Evaluate whether the new checkout flow reduces abandonment"
• Bad: "Understand the user experience" (too vague)
• Bad: "Explore the problem space" (no measurable outcome)

Write 3-5 research questions. More than 10 means your scope is too broad.

2. Select the Approach

See references/research-methods-catalog.md for detailed method descriptions.

3. Plan and Prepare

Every study needs a written plan, even if it's brief. Use templates/research-plan-template.md.

Essential plan elements:

• Problem statement and research questions
• Methodology and rationale
• Participant criteria and recruiting approach
• Number of participants (see benchmarks below)
• Timeline and logistics
• How data will be captured and stored
• Who will participate from the team

Participant benchmarks:

• Usability testing: 3-5 per round (catch ~85% of issues with 5 users)
• Generative interviews: 5-8 per persona segment
• Card sorting: 15-20 for statistical patterns
• Surveys: 100+ for quantitative significance
• Continuous discovery: 1 per week minimum

4. Recruit Participants

Recruiting is the hardest part. Embrace it — good recruiting puts the quality in your research.

A good participant:

• Shares the goals and concerns of target users
• Embodies key characteristics (age, role, tech familiarity)
• Can articulate their thoughts clearly
• Is familiar with relevant technology at the same level as target users

Write a screener survey (see templates/screener-template.md):

• Define specific behaviors you're looking for
• Determine required tool knowledge and access level
• Set topic knowledge requirements
• Don't make it obvious who you're looking for (people will game it)

Recruiting tips:

• The importance of recruiting perfectly representative users is overrated
• Testing early and often matters more than perfect representation
• Don't be embarrassed to ask friends and neighbors for initial rounds
• Offer reasonable incentives ($50-100/hour)
• Keep the invitation simple
• Don't discuss the product beforehand

5. Collect Data

Data management:

• Use consistent naming: Study-ParticipantName-YYYY-MM-DD
• Get files onto a shared drive immediately
• Use familiar tools — technical difficulties kill research sessions
• Keep a research kit ready to go at a moment's notice

Core collection methods:

Interviewing — The most effective way to see the world as users do. See references/interviewing-guide.md for the full technique guide.

Quick checklist:

• Prepare an interview guide with goals, demographics, icebreakers, and focus questions
• Ask open-ended questions; follow up with "Tell me more about that"
• Listen more than you speak — get comfortable with silence
• Note exact phrases and vocabulary participants use
• Pay attention after you stop recording — revelations often come then

Usability Testing — Observe representative users attempting tasks with your design. See references/usability-testing-guide.md for the complete process.

Quick checklist:

• Create tasks based on scenarios and personas
• The designer should NOT be the facilitator
• Have a separate observer taking notes
• Test early, test often — never right before lunch
• 3 users per round, debrief over lunch the same day

Contextual Inquiry — Interview and observe in the participant's actual environment.

Quick checklist:

• Plan for travel time
• Get situated without interrupting routines
• Ask people to do activities, not give you a tour
• Observe everything: environment, workarounds, tools, interruptions
• Use more than one researcher for multiple perspectives

6. Analyze Data

Analysis is where patterns become insights. Get everyone involved — include people who will benefit from participating.

Qualitative analysis process:

1. Review all notes and recordings closely
2. Extract observations: behaviors, emotions, actions, verbatim quotes
3. Write each observation on a sticky note (coded to the source participant)
4. Group notes into themes on a whiteboard
5. Watch patterns emerge; rearrange as you assess
6. Name each cluster with an insight statement
7. Derive actionable recommendations from insights

Analysis session structure:

• Summarize research goals and process (what did we want to find out?)
• Describe participants and circumstances
• Describe data gathering methods
• Pull out quotes and observations as a group
• Group into themes (e.g., "participants rely on pen and paper to aid memory")
• Summarize findings: patterns → insights → implications for design
• Document in a shareable format

See references/analysis-synthesis.md for detailed techniques including affinity diagrams, opportunity mapping, and persona creation.

7. Report Results

A quick sketch of a persona or a photo of sticky notes on a whiteboard is far superior to a lengthy written report that goes ignored.

Always include:

• Research goals and questions
• Methods and participant summary
• Key findings (patterns and insights)
• Severity-ranked recommendations
• Suggested next steps

Findings severity scale:

Findings frequency scale:

Issue triage (3 tiers):

• Tier 1: High severity + high frequency → immediate action, high risk to product success
• Tier 2: Moderate severity with low frequency, or low severity with moderate frequency
• Tier 3: Low severity + low frequency → low risk, add to backlog

Use templates/findings-report-template.md for the report structure.

Continuous Discovery

For teams practicing continuous discovery (recommended), research is not a phase — it's a weekly habit.

The continuous discovery cadence:

• Weekly touchpoints with customers by the team building the product
• Small research activities in pursuit of a desired outcome
• Infuse daily product decisions with continuous customer input

The product trio (product manager + designer + engineer) should:

• Conduct story-based interviews weekly (not opinion-based — ask about actual past behavior)
• Maintain an Opportunity Solution Tree mapping outcomes → opportunities → solutions → assumption tests
• Generate 15-20 solution ideas per opportunity (volume drives quality)
• Test 10-20 assumptions per week through small, rapid experiments
• Define success criteria before running tests

See references/continuous-discovery.md for the full framework.

Common Mistakes to Avoid

• Leading questions: "Don't you think this is easier?" → "How would you describe this experience?"
• Confirmation bias: Seeking evidence that supports your hypothesis instead of genuinely testing it
• Testing too late: Testing right before launch when there's no time to act on findings
• Asking opinions instead of observing behavior: People don't accurately predict their own behavior
• Over-documenting, under-acting: A 50-page report nobody reads vs. a whiteboard photo that drives change
• Skipping the screener: Recruiting anyone available instead of representative participants
• The facilitator is the designer: The person who built it can't objectively watch it fail
• Solving during analysis: Resist the urge to jump to solutions before the patterns are clear
• Adding features based on user requests: "I'd like it if it could do X" — probe deeper; they often already have a source for X

Quick Reference

Problem statement formula: [Verb: identify/evaluate/determine] + [what] + [for whom] + [context] Interview question format: Open-ended, behavior-focused, past-tense ("Tell me about the last time you...") Usability task format: Scenario-based, goal-oriented ("You want to [goal]. Show me how you'd do that.") Screener format: Behavioral questions that filter without revealing criteria Analysis output: Pattern → Insight → Recommendation → Severity

Resources

This skill includes:

• references/research-methods-catalog.md: Detailed guide to 20+ research methods organized by type
• references/interviewing-guide.md: Complete interviewing technique from preparation to debrief
• references/usability-testing-guide.md: Full testing process with scoring rubrics, triage, SUS (full 10-question scale with scoring formula, grade/percentile interpretation, subscales), and alternative questionnaires (UMUX-Lite, SEQ, NASA-TLX)
• references/analysis-synthesis.md: Affinity diagrams, thematic analysis, persona creation, opportunity mapping
• references/continuous-discovery.md: Weekly cadence, Opportunity Solution Trees, assumption testing
• references/workshop-methods.md: Team facilitation techniques — problem framing (HMW, Problem Tree, Abstraction Laddering, Rose/Thorn/Bud), ideation (Thumbnail Sketching, Creative Matrix, Round Robin, Alternative Worlds), prioritization (Importance/Difficulty Matrix, Bull's-Eye, Dot Voting, What's on Your Radar), and modeling (Storyboarding, Concept Poster, Cover Story Mock-Up). Includes half-day workshop template and facilitation anti-patterns.
• templates/research-plan-template.md: Fillable research plan
• templates/interview-guide-template.md: Interview script template
• templates/usability-test-plan-template.md: Test plan with tasks and metrics
• templates/screener-template.md: Participant screener survey
• templates/findings-report-template.md: Research findings report
• examples/method-selection-scenarios.md: "Given X situation, use Y method" decision guide
• examples/interview-to-insights.md: Walkthrough from raw interview data to actionable insights