Synthesize User Interview Notes into Counted Themes and Insights

Turns raw interview notes into a ranked theme table showing how many participants raised each theme, with observations kept separate from interpretation.

You getSummary, theme table with columns and contradictions and weak signals

Prompt

Prompt
You are a senior UX researcher who synthesizes interview data for product teams.

I will paste notes from several user interviews right after this message. If it is unclear where one participant's notes end and the next begin, ask me how to split them before you start.

Your job: group the notes into themes and pull out insights the team can act on.

Rules:
1. For every theme, count how many participants raised it (for example 4 of 7) and list their IDs. A participant counts once per theme.
2. Keep observations and interpretations separate. Observations are what participants said or did, tied to a participant ID. Interpretations are your reading of what it means, clearly labeled as such.
3. Use only what is in the notes. Do not invent quotes, numbers, behaviors or participants.
4. Flag contradictions where participants disagree, and mark any theme raised by only one person as a weak signal.
5. Rank themes by participant count, then by how severe the described problem is.

Output format:
1. Summary: 3 sentences on the biggest findings.
2. Theme table with columns: Theme | Participants (count and IDs) | Key observations | Interpretation | Confidence (High/Medium/Low)
3. Contradictions and weak signals: bulleted list.
4. Open questions to test next: 3-5 bullets.

End with one line on gaps in the notes or limits of the sample that could skew these findings.

Example output

At a glanceParticipants raising each theme (of 6)6 interviews with independent contractors about an invoicing app
  • Chasing late payments: 5 of 6
  • Invoicing from the job site: 4 of 6
  • Adding photos as proof of work: 3 of 6
  • Tax-time exports: 2 of 6
  • Team access for a helper: 1 of 6

Example for: 6 interviews with independent contractors about an invoicing app, notes labeled P1-P6.

1. Summary

Late payment chasing is the most common and most painful problem, raised by 5 of 6 participants. Most users invoice from their phone on job sites, but the mobile flow is where they get stuck. Pricing was rarely mentioned, which suggests it is not the main barrier to retention.

2. Theme Table

ThemeParticipantsKey observationsInterpretationConfidence
Chasing late payments5 of 6 (P1, P2, P3, P5, P6)P2 sends reminder texts by hand; P5 said "I spend Sunday nights chasing money"; P6 keeps a paper list of overdue clientsAutomated reminders may matter more than any new invoicing featureHigh
Invoicing from the job site4 of 6 (P1, P3, P4, P6)P3 writes invoices in the truck between jobs; P4 abandoned the mobile app twice when adding line itemsThe mobile line-item step is a likely drop-off pointHigh
Adding photos as proof of work3 of 6 (P1, P4, P5)P1 emails photos separately after invoicing; P5 had a client dispute a chargeAttaching photos could reduce disputes and speed up paymentMedium
Tax-time exports2 of 6 (P2, P6)P2 rebuilds a spreadsheet for her accountant each AprilReal need, but seasonal and smaller groupLow
Team access for a helper1 of 6 (P3)P3 wants his wife to send invoices without his loginWeak signal; may apply only to family-run businessesLow

3. Contradictions and Weak Signals

  • P4 wants fewer fields on invoices; P2 wants more detail for client records. The right default is unclear.
  • P1 likes reminder emails, but P5 worries automated reminders will annoy long-term clients.
  • Team access (P3 only) is a weak signal.

4. Open Questions to Test Next

  • Would users trust automatic reminders if they could edit the tone and timing?
  • Where exactly in the mobile line-item flow do users quit?
  • Do photo attachments change how fast clients pay?
  • Is the field-count disagreement linked to trade type or business size?

Gaps: All 6 participants are existing users in the trades, so churned users and other freelancer types are missing, and the counts reflect a small sample.

Written by AI from this prompt. Your results will vary.

Tips and common edits

Why this works

  • Counting participants per theme, with IDs, stops one loud interviewee from looking like a trend.
  • Splitting observations from interpretation lets the team challenge conclusions without losing the evidence.
  • Flagging contradictions and weak signals keeps the synthesis honest instead of tidy.

Use after a round of discovery or usability interviews. Paste all notes in one message in ChatGPT, Claude or Gemini, with each participant labeled P1, P2 and so on. Claude handles long transcripts well.

Common edits

  • Add a column for suggested product actions or opportunity size per theme.
  • Ask for a one-slide version of the summary to share with stakeholders.

Prompt details

By
Prompts-GPT editors
Published

Did this prompt work?

Keep exploring

More in Product Management