How do I start synthesizing messy interview notes?

Begin with a quick highlight pass for pains, outcomes, and workarounds. Tag quotes consistently, then affinity map tags into clusters. You're looking for repetition and tensions, not perfect phrasing.

How many interviews are enough to see patterns?

Typically 5–7 per segment reach thematic saturation. If new themes keep appearing, you have multiple segments or a scope issue. Document confidence and plan a quick survey to validate scale.

What's the difference between a theme and an insight?

Theme = repeated pattern (what happens). Insight = why it matters + implication. Good insights suggest a decision or a How‑Might‑We and connect to Jobs‑to‑Be‑Done.

How do I avoid cherry‑picking quotes?

Maintain a quotes database with counts per theme, include counter‑examples, and label source/participant. If possible, use a second coder on a subset and reconcile differences.

How does Jobs‑to‑Be‑Done fit into interview synthesis?

Map functional, emotional, and social jobs to your themes. Link quotes to jobs to explain behavior beyond feature requests and to drive solution framing.

What should the stakeholder readout include?

A one‑pager with top themes (with frequency), 2–3 killer quotes, HMW questions, and recommended next steps (prototype/measure/owner/date). Put raw notes and quotes in an appendix.

What do I do with strong outlier insights?

Treat them as hypotheses. Validate cheaply via mini survey, concierge test, or prototype. Don't let a single quote set the roadmap—capture and test it.

Should I create personas from this research round?

Only if behaviorally distinct clusters emerge that change decisions. Otherwise segment by task/context and keep personas lightweight to avoid stale artifacts.

How do I make insights actionable for engineering?

Translate to behaviors and constraints (e.g., drop‑offs, offline needs, trust issues), propose HMWs, and define prototype acceptance checks. Include the 2–3 quotes that humanize the problem.

User Interview Synthesis

Q: How do I prioritize opportunities from interviews?

Score by severity × frequency × business impact, note confidence, then rank with ICE/RICE. Outliers go to an "emerging signals" list until validated.

Analyze interview data and extract insights

researchintermediateAffinity MappingJobs-to-be-DoneThematic Analysis1000-1400 words

Customize Your Prompt

Fill in the variables to generate your personalized prompt

Interview Notes/Transcript

Additional Context (Optional)

Preview

See how your prompt will look with the current variables

You are a Senior UX Researcher synthesizing user interview data: [Interview Notes/Transcript]

Provide analysis using these frameworks:

## 1. KEY THEMES (Affinity Mapping)
### Theme 1: [Pattern Name]
- Frequency: [X of Y participants]
- Representative quotes: "[Quote 1]", "[Quote 2]"
- Implications: [What this means for the product]

### Theme 2: [Pattern Name]
[Repeat structure]

## 2. JOBS TO BE DONE
### Primary Job: [Main job]
- Functional aspect: [What they're trying to do]
- Emotional aspect: [How they want to feel]
- Social aspect: [How they want to be seen]
- Current solutions: [What they use now]
- Satisfaction level: [1-10 with reasoning]

## 3. PAIN POINTS & OPPORTUNITIES
| Pain Point | Severity | Frequency | Opportunity | Priority |
|------------|----------|-----------|-------------|----------|
| [Pain 1]   | High     | Daily     | [Solution]  | P0       |

## 4. USER QUOTES DATABASE
### Category: [Topic]
- "[Powerful quote 1]" - Participant A
- "[Powerful quote 2]" - Participant B

## 5. PERSONA REFINEMENTS
Based on interviews, update personas with:
- Behavioral patterns observed
- Mental models identified
- Decision criteria discovered
- Workflow variations

## 6. INSIGHTS & "HOW MIGHT WE" QUESTIONS
**Insight 1:** [Observation]
→ HMW: [How might we address this?]

**Insight 2:** [Observation]
→ HMW: [How might we address this?]

## 7. RECOMMENDED NEXT STEPS
1. **Validate:** [What needs quantitative validation]
2. **Prototype:** [What to test next]
3. **Research:** [Remaining questions]

Use thematic analysis and ensure insights are actionable for product development.

## Important Guidelines

### Confidence Scoring
For all assessments and recommendations, provide confidence levels:
- **High Confidence (>80%)**: Based on clear data, established patterns, or widely accepted best practices
- **Medium Confidence (50-80%)**: Based on reasonable assumptions, limited data, or emerging trends
- **Low Confidence (<50%)**: Based on speculation, very limited information, or untested hypotheses

### Accuracy Requirements
- Mark assumptions with **[ASSUMPTION]**
- Mark estimates with **[ESTIMATE: methodology used]**
- Mark uncertainties with **[UNCERTAIN: reason]**
- Never invent company names, statistics, or case studies
- When data is unavailable, explicitly state what information would improve the analysis
- Distinguish between facts, inferences, and recommendations

### Source Attribution
- General knowledge: "Based on industry standards..."
- Inferences: "This suggests that..."
- Speculation: "One possibility is..."
- Best practices: "Common approaches include..."

What Makes a Good Interview Synthesis

• Clear themes with evidence: frequency counts and representative quotes (not vibes).
• Links to Jobs‑to‑Be‑Done (functional, emotional, social) so decisions make sense.
• Traceability: each insight maps back to specific participants and notes.
• Actionability: "How Might We" questions and next steps—prototype, validate, measure.
• Prioritization: severity × frequency × business value with explicit confidence.

Common Interview Synthesis Mistakes

• Cherry‑picking memorable quotes without checking how often it happens.
• Vague themes (e.g., "Users want simplicity") with no product implication.
• Jumping to personas too early; ignore behaviorally distinct patterns.
• No counter‑examples—real data always has dissenting voices.
• Delivering a beautiful readout with zero next actions or owners.

Questions PMs Actually Ask (Interview Synthesis)

I have 10 messy transcripts. Where do I even start?

Timebox a first pass: highlight pain points, outcomes, and workarounds. Tag quotes, not opinions. Then affinity map: cluster tags until patterns emerge. You're looking for repetition and tension, not poetry.

How many interviews are "enough" before I see real patterns?

Usually 5–7 per segment to reach thematic saturation. If you're still hearing new problems at interview #8, you have multiple segments—or your script is too broad. Plan a quick survey to validate scale.

Theme vs insight—what's the difference?

Theme: a repeated pattern (what happens). Insight: the "so what" (why it matters + implication). Good insights produce a crisp HMW and a decision (prototype, measure, or punt).

How do I avoid confirmation bias when picking quotes?

Keep a quotes database with counts per theme, include counter‑quotes, and label confidence. If possible, do a blind second coder on a subset. Your goal is traceable, defensible calls—not a greatest hits album.

Where does Jobs‑to‑Be‑Done fit in?

Use JTBD as a lens to explain behavior: functional job, emotional relief, social signaling. Map key quotes to jobs. It turns "feature requests" into real progress customers are trying to make.

What do I hand off to stakeholders without a 30‑page deck?

A one‑pager: top 3–5 themes with frequency, 3 killer quotes, 2–3 HMWs, and a next‑step plan (prototype, metric, owner, date). Put the quotes database and notes in the appendix for the curious.

How do I prioritize opportunities from interviews?

Score by severity × frequency × business impact, then run ICE/RICE to rank. Outliers go on a "watchlist" until validated. Hot take: speed to learning beats perfect scoring.

We heard one wild outlier. Chase it or ignore it?

Neither. Park it in "emerging signals" and validate cheaply—mini survey, concierge test, or a prototype thread. Don't let one quote set your roadmap, but don't lose potential wedges either.

Should we create personas from this round?

Only if you see consistent behavioral clusters that affect decisions. Otherwise, use task/segment lenses and keep personas lightweight. Fake personas age like milk.

How do I make this useful for engineers?

Translate insights into behaviors and constraints: "Users abandon after 2nd step when docs are missing; need offline draft + autosave." Add HMWs, acceptance tests for prototypes, and the top 3 quotes that humanize the problem.

How to Use This Prompt