user-research-techniques（用户研究技术）

User Research Techniques

Guide to qualitative and quantitative research methods for understanding users, validating ideas, and informing product decisions.

When to Use This Skill

Auto-loaded by agents:

• research-ops - For research methods, planning, and best practices

Use when you need:

• Planning research studies
• Choosing research methods
• Conducting interviews or tests
• Analyzing research data
• Validating product decisions

Research Methods Matrix

           Quantitative <-- --> Qualitative
           (What & How Many)    (Why & How)
              |
Behavioral --+-- Analytics        Usability Testing
(What they  |    Surveys          Field Studies
  do)       |    A/B Tests        Diary Studies
              |
Attitudinal -+-- Surveys          Interviews
(What they  |    NPS              Focus Groups
  say)      |    Questionnaires   Concept Tests

Qualitative Methods

User Interviews

Understand problems, validate solutions, assess products. Use open-ended questions, listen more than talk (80/20 rule), and ask "why" 5 times. 5-8 participants per segment.

Usability Testing

Test product usability with moderated or unmoderated sessions. Recruit 5-8 participants, use think-aloud protocol, measure task completion, time on task, and satisfaction.

Field Studies

Observe users in their natural environment through contextual inquiry, shadowing, or diary studies. Best for understanding context and discovering workarounds.

Card Sorting

Understand mental models and information architecture. Open (users create categories), closed (sort into given categories), or hybrid.

Focus Groups

6-10 participant moderated discussions. Good for exploring opinions and generating ideas. Avoid for validation (groupthink risk).

Comprehensive guide: references/qualitative-methods-guide.md

Quantitative Methods

Surveys

Measure attitudes at scale with NPS (0-10), CSAT (1-5), CES (1-7), or custom questions. Keep short (<10 questions), avoid leading/double-barreled questions. 100+ respondents for directional insights, 384+ for statistical significance.

Analytics

Track behavioral data: engagement (DAU/WAU/MAU), conversion funnels, retention cohorts, feature adoption.

A/B Testing

Test variants with statistical rigor. Requires a clear hypothesis ("Changing CTA color from blue to green will increase click-through by 10%"), a single variable change, sufficient sample size (use an online calculator -- typically 1,000+ per variant for small effects), and a 1-2 week runtime. Only test when you have enough traffic; for early-stage products with <1,000 weekly visitors, use qualitative methods instead.

Key pitfalls: Stopping tests early when results look promising (peeking problem), testing too many variants at once (diluted significance), and ignoring segment effects (overall neutral but positive for one cohort).

Research Synthesis

Combine findings across studies to identify patterns. Three core approaches:

• Affinity mapping: Group observations by similarity, name the clusters, identify themes. Start here when you have raw interview notes.
• Thematic analysis: Code data into themes using a three-level system (codes → categories → themes). More rigorous than affinity mapping.
• Jobs-to-be-Done: Frame findings around the job users are hiring your product to do. Structure: "When [situation], I want to [motivation], so I can [outcome]."

For deeper synthesis methods, see synthesis-frameworks.

Comprehensive guide: references/quantitative-methods-guide.md

Best Practices

1. Avoid Bias

Confirmation Bias: Seek disconfirming evidence Leading Questions: Ask neutral questions Selection Bias: Recruit diverse participants Observer Effect: Users behave differently when watched

2. Sample Sizes

Qualitative:

• 5-8 users per segment (diminishing returns)
• 15-20 total for diverse product

Quantitative:

• 100+ for trends
• 384+ for statistical significance
• Use power calculations

3. Triangulate

Combine Methods:

• Interviews (why) + Analytics (what)
• Usability tests + Surveys
• Quantitative -> Qualitative -> Quantitative

4. Continuous Discovery (Teresa Torres)

Weekly Touchpoints:

• Talk to 2-3 customers per week
• Mix research types
• Share with team
• Document insights
• Map to opportunities

Common Mistakes

Avoid:

• Asking what users want (they don't know)
• Leading questions ("Do you love this?")
• Only talking to power users
• Research without action
• Skipping synthesis

Do:

• Observe behavior, not just opinions
• Ask open-ended questions
• Recruit diverse participants
• Act on findings
• Share insights widely

Tools

Research Platforms:

• UserTesting, Maze (unmoderated testing)
• User Interviews, Respondent.io (recruitment)
• Lookback, Zoom (moderated testing)

Analysis:

• Dovetail, Airtable (synthesis)
• Miro, FigJam (affinity mapping)
• Typeform, SurveyMonkey (surveys)

Analytics:

• Mixpanel, Amplitude (product analytics)
• Hotjar, FullStory (session replay)
• Google Analytics (web analytics)

Templates and References

Assets (Ready-to-Use)

• assets/research-plan-template.md - Research plan template with goals, methods, questions, and deliverables

References (Deep Dives)

• references/qualitative-methods-guide.md - User interviews, usability testing, field studies, card sorting, focus groups
• references/quantitative-methods-guide.md - Surveys, analytics, A/B testing, research synthesis methods
• references/research-planning-guide.md - Defining research questions, choosing methods, recruiting participants

Resources

Books:

• "The Mom Test" - Rob Fitzpatrick
• "Just Enough Research" - Erika Hall
• "Continuous Discovery Habits" - Teresa Torres
• "Don't Make Me Think" - Steve Krug

Online:

• Nielsen Norman Group articles
• IDEO Design Kit
• Google Ventures Research Sprint

Quick Guide

Need to understand why? -> Interviews
Testing usability? -> Usability Tests
Measure satisfaction? -> Survey (NPS/CSAT)
Understand behavior? -> Analytics
Validate solution? -> Prototype Test
Deep context? -> Field Study

Always: Define questions, recruit right users, synthesize, act on insights

Troubleshooting

"I don't know which research method to use": Start with your question type. Exploring "why" = interviews. Testing usability = usability test. Measuring satisfaction = survey. If you're unsure, default to 5 user interviews -- they're the highest insight-per-hour method.

"I can't recruit enough participants": Lower the bar. For qualitative research, 5 is enough. Use your existing users, social media, or offer small incentives ($25 gift cards). For guerrilla testing, find people at coffee shops or coworking spaces.

"Leadership wants data but we have no budget for research": Use free methods: 5-second tests (UsabilityHub free tier), unmoderated testing (Maze free plan), or analyze existing support tickets and app reviews as proxy research data.

Related Skills

• interview-frameworks - Deep-dive interview techniques and question design
• usability-frameworks - Usability testing methodology and heuristic evaluation
• synthesis-frameworks - Turning research data into actionable insights
• validation-frameworks - Experiment design for validating assumptions