Customer Feedback Synthesizer

You are manually pulling reviews from G2, Trustpilot, App Store comments, and Reddit threads into spreadsheets every week. You skim hundreds of comments, tag themes by hand, and still miss what is actually driving churn, expansion, or support volume. By the time your summary reaches product or leadership, the signal is stale and the action list is vague. The fix is not another "overall sentiment" chart. Customer Feedback Synthesizer is a Support skill that ingests multi-source feedback exports, normalizes language, scores sentiment by source, clusters recurring themes, tracks NPS movement over time, and ranks feature requests by frequency x sentiment weight x account impact. It produces decision-ready outputs your product and support teams can use immediately. Anti-pattern it refuses: the "Single Average Score" shortcut. This skill will not collapse all channels into one blended sentiment number, because source mix distortion hides high-risk issues (for example, Reddit complaint spikes masked by stable app-store ratings). Instead, it keeps source-level weighting and trend breaks explicit. What you get: SKILL.md - operating protocol and run sequence. FEEDBACK_INGEST_WORKFLOW.md - source export and normalization steps. SENTIMENT_SCORING_RULES.md - channel-aware scoring logic. THEME_CLUSTER_TEMPLATE.csv - reusable clustering schema. NPS_TREND_ANALYSIS.md - rolling cohort trend method. FEATURE_PRIORITY_MATRIX.csv - weighted prioritization model. Requires CSV/JSON exports from review platforms, optional Reddit API key, and a spreadsheet or BI tool.