VoC Cluster Report
Question it answers: What themes, topics, and pain points do our customers actually talk about most — and how is that shifting over time?
Cadence: Weekly
GTM owner: PMM
Data sources
Internal: Conversation clusters from the V2 pipeline (cluster_search, cluster_detail). Each cluster represents a recurring topic across all customer interactions — tagged with label, summary, member count, top quotes, sentiment, and trending direction.
External: None. This is purely an internal voice read.
Living Document
A weekly KB document produced by a blueprint run. It reads the top 10–15 clusters ranked by member count and recency, then synthesizes a narrative report covering:
- Top themes this week — each cluster gets a 2–3 sentence summary grounding it in representative quotes from actual calls
- Rising vs. fading topics — what's appearing more frequently compared to the prior week
- Sentiment breakdown — which themes carry mostly positive, negative, or mixed sentiment
- Implications for PMM — where messaging is landing vs. where customers are expressing confusion or unmet needs
- One quote per theme — a pull quote from a real call conversation
The document is versioned by week. Each new run creates a proposed version; a PMM or content lead promotes it to current after review. Older versions stay accessible for trend comparison.
Page
A live in-console view for anyone who wants the cluster landscape without reading a narrative.
Stats row:
- Total active clusters this period
- Avg sentiment score across all customer interactions
- % of interactions with a tagged cluster
Bar chart: Cluster size by label (member count), top 15, sorted descending.
Line chart: Trend of top 5 cluster sizes over the last 8 weeks — shows which themes are growing or shrinking.
Table: One row per cluster — label, member count, top topic tags, avg sentiment, example quote, week-over-week change.
Users can click a cluster row to drill into cluster_detail — the underlying interactions and representative quotes for that theme.