You get:
- building features no one asked for while ignoring real requests
- surprise churn from problems you could have seen coming
- sales objections you could have pre-empted
- competitors winning on things customers complain about in your reviews
- product managers disconnected from user pain
But reviews reveal patterns:
- what customers love (your differentiators, accidentally)
- what they hate (churn drivers, fix these first)
- what they beg for (feature gaps, prioritize these)
- who they compare you to (unexpected competitors)
- pricing feedback (too expensive? too cheap?)
Without synthesis, you’re drowning in data but starving for insight.
This prompt turns raw review text into actionable product intelligence.
Assume the role of a product analyst who synthesizes customer reviews. Your task is to aggregate and analyze review text from multiple sources. Generate: 1. SENTIMENT SUMMARY - Overall star rating distribution - Sentiment trend (improving/declining/stable) 2. TOP 5 LOVES (with verbatim examples) - What customers praise most - Specific features mentioned - Unexpected delights 3. TOP 5 HATES (with verbatim examples) - What customers complain about most - Recurring bugs or usability issues - Missing features they expected 4. TOP 5 REQUESTS (feature gaps) - "I wish it had X" - "If only Y worked" - "Compared to Z, you're missing..." 5. COMPETITOR MENTIONS - Which competitors are mentioned - Why customers switched (to or from) 6. SEGMENT DIFFERENCES - Small business vs. enterprise (different loves/hates) - New users vs. power users - Industry-specific feedback 7. URGENT ACTION ITEMS (ranked by customer impact) - Bugs to fix immediately - Features to build next - Documentation to add INPUTS: Review source 1 (G2 / Capterra / Trustpilot): [PASTE 10-20 REVIEWS OR SUMMARY] Review source 2: [PASTE] Review source 3: [PASTE] Your product category: [E.G., "Project management software"] Time period (if known): [E.G., "Last 6 months"] RULES: - Use direct customer quotes wherever possible - Flag reviews that mention pricing (emotional signal) - Identify "silent majority" themes (not just loud outliers) - Note when the same complaint appears across competitors (industry problem)
- Export 20-50 recent reviews from G2, Capterra, or Trustpilot — more is better.
- Run this monthly for your own product, quarterly for key competitors.
- Pay special attention to 3-star reviews — they’re often the most thoughtful.
- Look for requests that appear in both your reviews and competitor reviews (industry-wide gap).
- Share the “top hates” with engineering for immediate bug triage.
Review source 1 (G2):
“5 stars: Love the mobile app. So fast. Finally a project tool that doesn’t crash.
4 stars: Good but reporting is weak. Can’t export to Excel.
2 stars: Support took 3 days to respond. Lost a client because of it.
3 stars: It’s fine. Slack integration is buggy — notifications don’t always show.”
Review source 2 (Capterra):
“4 stars: Great for small teams. Price is right. But no Gantt charts.
1 star: Absolute nightmare. Lost all my data after an update. Support unreachable.
5 stars: The timeline view is brilliant. Beats Asana for my use case.”
Your product category:
Project management software
This framework improves outcomes by forcing:
- sentiment aggregation (not just anecdotes)
- verbatim quotes (customer voice preserved)
- competitor mention tracking (who you’re losing to)
- segment differences (one fix doesn’t fit all)
- ranked action items (prioritization, not list)
Great review synthesis doesn’t just summarize — it tells you what to do next.
Build Better AI Systems
Subscribe for advanced prompt engineering, AI coding tools, debugging frameworks, and practical strategies for developers and engineers.