Product Discovery 14 min
Feedback Synthesis
Cluster customer feedback into themes without losing evidence.
Learning Objectives
- Cluster qualitative feedback.
- Preserve quotes and source context.
- Assign confidence to product recommendations.
Preserve the customer's words
AI can cluster customer feedback quickly, but synthesis becomes dangerous when it loses evidence. Keep quotes attached to every theme.
Ask for theme, evidence quotes, segment, suggested action, and confidence. Confidence matters because feedback is often ambiguous. The model should not treat one loud comment like a market signal.
Better synthesis prompts
- Use only provided quotes.
- Group similar problems, not similar words.
- Name the user segment when available.
- Separate customer pain from product solution.
- Include confidence and open questions.
Good synthesis helps a product team decide what to investigate, build, ignore, or clarify.
Examples
Evidence-preserving themes
Theme: onboarding confusion. Evidence: three quotes from admins. Action: simplify checklist. Confidence: medium.
Practice Exercise
Cluster feedback
Paste a batch of feedback and synthesize it into themes with evidence quotes.
- Every theme has quotes.
- Suggested actions are tied to evidence.
- Confidence is included.
Mini Prompt Templates
Feedback Synthesizer
Feedback: [CUSTOMER_FEEDBACK] Task: Cluster feedback into themes and recommend product actions. Format: Theme, Evidence Quotes, Segment, Suggested Action, Confidence. Constraints: Use only provided quotes.