Turning raw interview notes into themes you can act on
Synthesis is the single most common AI task for PMs: about half use AI to make sense of research and feedback faster. The bottleneck is real, a dozen customer calls produce pages of messy notes and a deadline, but so is the failure mode, models invent quotes and overstate patterns. The fix is structural: the AI organizes and counts, using only the text you paste, and never fabricates a verbatim.
You are a user research analyst helping a product manager synthesize customer interviews. Work only from the notes I paste; do not add outside knowledge. Research question: {{research_question}} Product and user context: {{product_context}} Interview notes (multiple participants, separated by "---"): {{research_notes}} Produce: 1. A themes table with columns: theme, one representative verbatim quote copied exactly from the notes, number of participants who raised it, severity (blocker / friction / nice-to-have), and the underlying job-to-be-done or opportunity. 2. A short "signal vs. noise" note flagging which themes came from only one participant, so I don't over-index on a single voice. 3. Open questions the notes don't answer, each written as "Not covered in these interviews — investigate [topic]". Rules: - Every quote must appear verbatim in the notes I pasted. Never paraphrase into quotation marks and never invent a quote. If you cannot find a real quote for a theme, write "[no direct quote]". - Do not infer sentiment, feature requests, or willingness to pay that participants did not actually state. - Order themes by number of participants, then severity. Keep the whole output under 500 words.
Fill in your details and the prompt updates live — then copy.
| Theme | Quote | # Participants | Severity | Opportunity | |---|---|---|---|---| | Unclear which integration to start with | "I couldn't tell which integration to pick" | 4 of 6 | Blocker | Guided first-source picker | | Setup feels heavy before any value | "It felt like a lot before I saw anything" | 3 of 6 | Friction | Show a sample dashboard pre-connect | Signal vs. noise: "Wants a mobile app" came from P5 only — single voice, do not roadmap on this alone. Open questions: Not covered in these interviews — investigate whether drop-off differs by company size.
The full workflow
- Strip names, company names, and any personal identifiers from the notes before pasting
- Confirm your transcription tool and AI account are company-approved for customer data
- Run the prompt, then spot-check three or four quotes against the source transcript
- Pull the flagged single-voice themes out of your headline findings
- Bring the themes table to the team as evidence, not as a conclusion
Watch out for
Interview recordings and transcripts contain personal data. Under GDPR and CCPA you need a lawful basis and, often, participant consent to process them; a consumer AI account that trains on inputs is not an approved processor. Use an enterprise tool with training and retention disabled, and anonymize before pasting.
Models fabricate quotes that sound exactly like your users. Verify every verbatim against the transcript before it lands in a deck, or you will present words no one said.
Frequency is not importance. Four people mentioning something in six calls is a signal to explore, not a validated priority.
Where this comes from
Every use case on this site is grounded in real reports from working product managers — not invented by us.