alexe-ev Synthesize Feedback Themes
Purpose
Help teams turn high-volume, unstructured customer feedback (surveys, support tickets, reviews, interviews) into a structured set of themes that can be prioritized and acted on.
Skill type
Conceptual skill
Use this skill when
- A large volume of feedback exists but hasn't been organized into patterns
- Customer support, surveys, and review data need to be unified
- A feedback synthesis is needed before a roadmap planning cycle
- Product decisions need to be grounded in customer voice
Do not use this skill when
- Feedback hasn't been collected yet (collect it first)
- The goal is individual account management (this is population-level synthesis)
Required inputs
- Feedback sources and rough volume
- Synthesis goal: what decisions will this inform?
Optional inputs
- Raw feedback data or samples
- Segment breakdown of feedback sources
- Prior synthesis for comparison
- Product area focus
Upstream context
Works best when:
- Multiple feedback channels are active
- VoC program is running
Downstream handoff
Output can feed:
- triage-feedback-loop (themes → triage and routing decisions)
- identify-problem-opportunity (themes → opportunity framing)
- build-roadmap-prioritization (themes → roadmap inputs)
Instructions
- Collect all feedback sources: support tickets, survey responses, review sites, interviews, NPS verbatims.
- Read a sample across all sources to calibrate.
- Code feedback by theme: functional issues, emotional signals, feature requests, praise, competitive mentions.
- Quantify themes: what percentage of feedback touches each theme?
- Segment by customer type if data allows.
- Identify the top 5–8 themes with supporting evidence.
- Separate what customers say from what they need (interpret, don't just quote).
Output
Provide:
- Feedback source inventory (channels, volume, date range)
- Theme set (5–8 themes)
- Each theme: description, frequency/volume, representative quotes, customer segment breakdown
- Top themes by volume
- Themes that are high-emotion (even if lower volume)
- Implicit needs (what customers are expressing that they haven't said directly)
- Recommended actions per theme
Risks / caveats
- Loudest feedback is not most representative — vocal minority bias is real
- Feature requests are often symptoms of deeper problems — interpret the need, not the request
- Synthesis without segment breakdown obscures important differences between customer types