UX Research

UX Research

Plan and execute user research that produces decisions, not just decks. Stack-agnostic. Tool-agnostic.

This skill is for generative and discovery research. For testing existing designs, use usability-testing. For mapping the full customer experience, use journey-mapping.

---

When to use

• Starting a new product or major feature without sufficient user understanding
• Diagnosing why something isn't working without clear data signals
• Generating new opportunity hypotheses
• Validating a strategic direction before significant investment
• Building empathy across a team that's drifted from users
• Translating "we should talk to users" intent into a real plan

When NOT to use

• Testing a specific design or prototype (use usability-testing)
• Mapping the full journey of an existing experience (use journey-mapping)
• Quantitative measurement (use analytics-strategy)
• Conversion testing (use cro-optimization)

---

Required inputs

• The research question(s) - what you need to answer
• Stakeholder buy-in (who needs the findings, what decisions hinge on them)
• Access to users (current customers, prospects, lapsed users, target segments)
• Timeline and budget
• Any prior research to build on

---

The framework: 6 phases

1. Frame the question

Bad questions produce bad research. Spend disproportionate time on framing.

Good research questions:

• Specific (not "How do users feel about our product?")
• Open-ended (not "Do users like feature X?")
• Decision-relevant (the answer changes what gets built)
• Researchable (can be answered through user contact, not just analysis)

Examples:

2. Choose the method

The method follows the question.

Generative methods (what's true?):

• In-depth interviews. 60 minutes, 5 to 15 participants. Best for understanding context, motivation, mental models.
• Contextual inquiry. Observe users in their environment doing their work. Best for workflow understanding.
• Diary studies. Participants log their experience over days/weeks. Best for behaviors that don't manifest in a single session.
• Field research. Spend time where users live/work. Best for cultural and contextual understanding.
• Surveys (qualitative-heavy). When you need broad signal with open-ended responses.

Validation methods (is this hypothesis right?):

• Concept testing. Show a description, mockup, or prototype. Get reactions.
• Card sorts. Validate information architecture.
• Tree tests. Validate findability without visual design influence.

(For testing usability of working designs, see usability-testing.)

3. Recruit

The recruit makes or breaks the research.

Recruit criteria:

• Match the audience the research targets (not "anyone willing")
• Mix of behaviors (active users, lapsed users, never-users)
• Mix of demographics where relevant
• Excludes friends, family, employees (biased)
• Excludes professional research participants if possible (different population)

Recruit channels:

• In-product recruiting (intercept current users)
• Email outreach to user segments
• Recruiting platforms (UserInterviews, Respondent, etc.)
• Customer support team referrals
• Field intercept for in-person

Incentive: Pay participants. Standard rates: $50 to $150 for 60 minutes, more for executives or specialized professions.

Recruit volume: Plan for 20 to 30 percent no-show. Recruit 7 to schedule 5.

4. Conduct

The interview or session itself.

Pre-interview:

• Send confirmation 24 hours and 1 hour before
• Test recording setup (audio quality is non-negotiable)
• Prepare interview guide (see template)
• Have a notetaker if possible (frees the interviewer to focus)

During the interview:

• Record video and audio (with consent)
• Open with rapport-building, not the research questions
• Use open-ended questions ("Tell me about the last time...")
• Use silence (let participants fill it; don't rush to the next question)
• Ask "why" but not too many times in a row (becomes interrogation)
• Ask for specifics and examples ("Can you walk me through what you did?")
• Probe contradictions gently ("Earlier you said X, now you're saying Y; help me understand")
• Watch for moments of emotion (often signal something important)
• Don't sell or convince - this is listening, not pitching

Anti-patterns:

• Leading questions ("Don't you find this confusing?")
• Hypothetical questions ("Would you use a feature that...?") - poor predictor of behavior
• Multiple questions at once
• Interrupting
• Filling silence
• Interviewing your hypothesis (only asking questions that confirm what you already think)

5. Synthesize

Notes don't become insights automatically.

The synthesis process:

1. Capture observations. From recordings, notes, transcripts. Each observation is a single data point: a quote, a behavior, an emotion, a moment.
2. Affinity mapping. Cluster observations into themes. Physical sticky notes or digital equivalents.
3. Find patterns. Themes that appear across multiple participants are signal. One-off observations are interesting but weaker.
4. Identify insights. An insight is more than a theme. It's a non-obvious finding that explains a why or implies a so what.
5. Test the insight against the data. If the insight only fits some interviews, it's a hypothesis, not an insight.
6. Distinguish signal from noise. A belief that 1 of 8 participants holds may be noise. A belief 6 of 8 hold is signal.

Heuristics for strong insights:

• They surprise the team (insights you already knew aren't insights)
• They explain a "why" the team has been guessing about
• They imply specific actions (so what?)
• They hold up across multiple data points
• They can be stated in one or two sentences