Summarize 5 Customer Interviews Into 3 Insights
Distills a round of 5 customer interviews into 3 prioritized insights for leadership. Built for PMs who need to compress rich qualitative data into exec-ready findings.
When to use this prompt
Use this after completing a round of 5 or more customer interviews when you need to share findings with leadership. You will need the interview notes or transcripts and the original research question. The prompt deliberately limits the output to exactly 3 insights because leadership cannot act on more than that in a single review. It forces you to choose what matters most, which is the hardest part of research synthesis. Do not use this to summarize 1-2 interviews; there is not enough signal to draw insights. Use it when you have genuine pattern recognition across multiple customers.
The Prompt
You are a product manager synthesizing 5 customer interviews into exactly 3 prioritized insights for a leadership audience. The rule of 3 is strict: no more, no fewer. Research question: {{research_question}} Interview notes: {{interview_notes}} Interviewee roles: {{interviewee_roles}} Strategic context: {{strategic_context}} Produce the synthesis with this structure: 1. RESEARCH RECAP (2 sentences) â Restate the question and the interviewee composition. 2. INSIGHT 1 â Title, 2-sentence summary, 3 supporting quotes (with attribution), estimated impact, recommended action. 3. INSIGHT 2 â Same structure. 4. INSIGHT 3 â Same structure. 5. WHAT WE DID NOT FIND â 2-3 sentences on hypotheses we went in with that were NOT confirmed. This is often as valuable as what we did find. 6. NEXT RESEARCH STEP â 1-2 sentences on what should be validated next, either with more interviews or quantitative data. Insight quality rules: - Each insight must be supported by at least 3 of the 5 interviewees. Single-interview insights are anecdotes, not insights. - Each insight must be actionable. "Customers want better onboarding" is not actionable; "customers drop off specifically at the invite step because they cannot find teammate emails" is actionable. - Quotes must be verbatim, short (under 20 words), and attributed by role and interview number (e.g., "Interview 3 - VP Sales"). - Recommended actions should be specific enough to put in a backlog. If the interviews do not support 3 distinct insights, say so honestly. Do not pad with weak insights. Two strong insights are better than three mediocre ones.
Example Output
RESEARCH RECAP: We interviewed 5 billing admins at mid-market accounts to understand how they manage overdue invoices. Interviewees ranged from 3 to 12 years of experience. INSIGHT 1: Billing admins build personal spreadsheet systems because the product does not surface what they need. Summary: All 5 admins maintain their own Excel files to track overdue status, copying data from our product each morning. The duplication is a workflow crutch, not a preference. - "I rebuild the same view every day from CSV" (Interview 1, Finance Lead). - "I do not trust your numbers so I keep my own" (Interview 3, AP Manager). - "Spreadsheet is faster than your filter" (Interview 4, Controller). Impact: Losing 20 minutes daily per admin; trust erosion. Action: Ship overdue-first dashboard sort as top priority. INSIGHT 2: Admins do not chase invoices proactively; they wait for reminders from customers. Summary: 4 of 5 admins admit they only chase overdue invoices when their CFO asks, not on a schedule. This surprised us. - "I only chase when my boss asks" (Interview 2). - "We chase at month-end, not daily" (Interview 5). Impact: Automated reminders may be less valuable than we assumed because admins are reactive, not proactive. Action: Pause automated reminder exploration; pivot to month-end chase assist instead. INSIGHT 3: Trust in data accuracy is a higher priority than automation. Summary: 5 of 5 mentioned concerns about data freshness before mentioning any automation request. Fix trust before adding features. - "I need to know the data is live" (Interview 1). - "If the number is wrong I look bad" (Interview 3). Action: Add visible "last updated" timestamp; audit data freshness SLOs. WHAT WE DID NOT FIND: No admin mentioned wanting SMS notifications, despite our hypothesis. None prioritized multi-currency support (a design team assumption). NEXT RESEARCH STEP: Quantitative validation of the "trust before automation" signal across 50 accounts via a targeted survey.
Recommended Tools
Dovetail is purpose-built for interview synthesis and preserves transcripts, timestamps, and quote provenance, which matters when execs push back on a claim. Productboard links insights to roadmap decisions downstream so the research trail is visible in planning. Notion projects works for lighter teams without a dedicated research tool. Use Dovetail if research is a recurring practice; Notion if it is ad hoc.
Frequently Asked Questions
When should I use this prompt?
Use it after finishing a round of 5-10 customer interviews and before presenting findings to leadership. The rule of 3 is essential because execs cannot act on more than 3 findings in one review, and more findings dilutes the signal. If you have fewer than 5 interviews, use a less structured summary; you do not have enough pattern recognition across customers to claim insights. Also skip it for usability tests or moderated prototype reviews; those need different synthesis approaches focused on task success and observed behavior.
What if I have more than 3 important insights?
Pick the 3 most actionable and hold the rest for the next review. Forcing 5 insights into a single leadership review almost always causes 3 of them to be forgotten within a week. Leadership attention is the scarce resource, not insight supply. The discipline of picking 3 also improves your own thinking: if you cannot prioritize, you are not yet done synthesizing. The insights you hold back should be documented separately and revisited when the current batch has been acted on; they are not lost, just sequenced.