Synthesize a stack of user interviews into themes
Turn a pile of interview transcripts into ranked themes backed by verbatim quotes — the synthesis step that usually eats a day, done in an afternoon and traceable back to the source.
What you’ll get
Ranked themes across multiple interviews, each backed by real quotes with attribution — not a vague summary you can’t trust. The guardrail that makes this safe is forcing every claim to carry a verbatim quote, so you can catch hallucination instantly.
Before you start
Get clean transcripts (see Otter.ai for capture). Label each one so quotes stay attributable — P1, P2, etc. Claude’s long context comfortably holds several transcripts at once; if you have many, batch them in groups of five.
The prompt
You are a user researcher synthesizing interviews. Below are <N> transcripts,
each labeled P1, P2, …
Task:
1. Identify the recurring themes across participants (not per-interview summaries).
2. Rank themes by how many distinct participants raised them.
3. For EACH theme, include:
- a one-line description
- the participants who raised it (e.g. P1, P4, P7)
- 2–3 VERBATIM quotes, each tagged with the participant
4. Separately, list surprises — things only one person said that felt important.
Hard rule: every quote must be word-for-word from the transcripts. If you can't
find a real quote for a theme, say so instead of paraphrasing.
Transcripts:
"""
<paste labeled transcripts>
"""Then do the human part
- Spot-check three quotes against the source transcripts. If they’re accurate, trust the pattern; if any are paraphrased, re-run with a stronger warning.
- Take the top 2–3 themes straight into your PRD’s “Problem & context” as evidence, quotes included — traceable from decision back to the person who said it.