User Interview Synthesizer

Turn raw interview notes, call transcripts, or usability test observations into structured, reusable insights. Designed for consumer/mobile PMs who talk to users frequently and need to spot patterns fast.

When to Use

• "Synthesize these interview notes"
• "What themes are in these user calls?"
• "Process my research notes"
• "Summarize this usability test"
• After a batch of customer calls or research sessions
• When building a case for a feature or pivot

Process

1. Ingest the Raw Input

Accept notes in any format — bullet points, transcripts, pasted call notes, voice memo transcriptions. Ask:

• How many sessions? (1 interview or a batch)
• Who were the participants? (persona, segment, new vs. power user)
• What were you trying to learn? (the research question)

2. Extract Per-Session

For each interview/session, pull out:

Quotes — verbatim user language (these are gold for stakeholder buy-in)

• Flag emotional language ("I hate when...", "I wish...", "I love that...")
• Note workarounds users have built (signals unmet need)

Pain Points — problems the user described or demonstrated

• Severity: blocks a task vs. annoying vs. cosmetic
• Frequency: every time vs. sometimes vs. once

Goals & Motivations — what the user is ultimately trying to accomplish

• Functional goals (get X done)
• Emotional goals (feel confident, feel in control)
• Social goals (look good to others, share with friends)

Behaviors — what the user actually does (vs. what they say they do)

• Current workflow and tools
• Hacks, workarounds, alternative products used

Surprises — anything that challenged your assumptions

3. Cross-Session Synthesis

If multiple sessions, identify:

Themes — patterns that appear across 2+ sessions

• Rank by frequency (how many users mentioned it)
• Note convergence vs. divergence across segments

Jobs to Be Done — cluster findings into JTBD format

• "When I [situation], I want to [motivation], so I can [outcome]"
• Reference knowledge/customers/ for existing persona alignment

Opportunity Areas — gaps between user goals and current product

• Map to existing roadmap items if applicable (knowledge/products/*/roadmap.md)

4. Assess Confidence

For each theme or insight:

• Strong signal — 3+ users, consistent, unprompted
• Emerging signal — 2 users, or prompted but consistent
• Weak signal — 1 user, or contradicted by others

5. Store Insights

Write findings to memory/feedback/YYYY-MM-DD_[topic].md for future reference. Update knowledge/customers/ personas if new information warrants it.

Output

## Research Synthesis: [Topic]

**Date:** YYYY-MM-DD
**Sessions:** [count]
**Participants:** [persona/segment breakdown]
**Research question:** [what we wanted to learn]

### Top Themes
| # | Theme | Frequency | Confidence | Key Quote |
|---|-------|-----------|------------|-----------|
| 1 | [theme] | [X of Y users] | Strong/Emerging/Weak | "[quote]" |

### Pain Points
| Pain Point | Severity | Frequency | Segment |
|------------|----------|-----------|---------|
| [pain] | High/Med/Low | [X of Y] | [segment] |

### Jobs to Be Done
- When I [situation], I want to [motivation], so I can [outcome]

### Opportunity Areas
| Opportunity | Roadmap Link | User Evidence |
|-------------|-------------|---------------|
| [opportunity] | [existing item or "new"] | [summary] |

### Surprises & Assumptions Challenged
- [surprise]

### Recommended Next Steps
- [action: more research / build experiment / write PRD / update roadmap]

Principles

• Verbatim quotes > your interpretation — capture their words
• Behaviors > opinions — what they do matters more than what they say
• Patterns > anecdotes — one user's rant isn't a trend
• Connect to roadmap — insights are useless if they don't influence decisions