Voice of Customer Synthesis
Turn the noise of many signals into a few ranked problems. This is the step before a PRD: /prd takes the one signal that matters; this skill finds it out of many. The value is the synthesis judgment — cluster by underlying problem, score by frequency × severity × strategic fit, and refuse to let the loudest theme automatically win.
If fewer than 5 signals are provided, say so and recommend writing a PRD on the strongest one directly. If none are provided, ask for them — never invent signals.
Required structure
- Snapshot —
N signals → M themes, the date, and the source mix.
- Theme table (sorted by priority, not frequency): Theme (the underlying problem) | Signals (n, %) | Severity | Fit | Priority | Representative quote.
- Severity = depth of pain per signal (High = blocks the job / costs money / drives churn). Fit = alignment with product direction. Priority = a transparent roll-up you can explain, never a black box.
- Every theme cites a verbatim quote. No quote, no theme.
- Prioritization read — loudest ≠ most important — is the most-frequent theme also the highest-priority? Name the squeaky wheel (loud, low-severity) and the silent killer (rare, account-ending).
- Top 3 problems (PRD-ready) — ranked. Each: underlying problem, segment, business cost, a verbatim quote, and a copy-paste
/prd "..." line.
- What I did NOT hear — sampling bias and coverage gaps. Never pretend the sample is the population.
- Recommended next step — which problem to take to a PRD first, and why.
Hard rules
- Work only from the provided signals — never invent quotes, inflate counts, or fabricate segment size / revenue.
- Cluster by underlying problem, not by feature or product area. 3–7 themes; more than ~8 means you're clustering by surface.
- Show the frequency × severity × fit scoring transparently.
- Always deliver the "loudest ≠ most important" read and flag any silent killer.
- Always flag sampling bias.
- Top-3 must be PRD-ready, not vague theme labels.
- Active voice. State confidence instead of hedging.
When a dedicated tool fits better
For hundreds of signals, continuous ingestion, multi-PM tagging, or governed verbatims, a purpose-built VoC platform (Dovetail, Viable, Productboard Insights, Thematic) is the right call — then bring its top themes here for the prioritization read and PRD handoff. This skill's edge is zero setup, no data egress, and the clean handoff to the next step.
Where to write
- With filesystem:
outputs/voc-<topic-slug>-<YYYY-MM-DD>.md (slug names the source/area, not the winning problem).
- Inline (web/mobile): emit as markdown with the suggested filename at the top.
Tone
Direct, customer-centric, quantified where the signals allow, honest about the sample where they don't. Lead with the problem, not the feature.
1---2name: pmstack-voc3description: Synthesize many raw customer signals (support tickets, interview snippets, churn reasons, sales notes, NPS verbatims, app reviews) into a small set of ranked, PRD-ready problems. Use when a PM pastes or attaches a pile of feedback and asks "what's the real problem / what should we build first / what are the themes", when there are too many signals to read one by one, or when someone needs to turn voice-of-customer data into prioritized problem statements before writing a PRD. This is the front-of-funnel step that precedes pmstack-prd.4---56# Voice of Customer Synthesis78Turn the noise of many signals into a few ranked problems. This is the step *before* a PRD: `/prd` takes the one signal that matters; this skill finds it out of many. The value is the synthesis judgment — cluster by underlying problem, score by **frequency × severity × strategic fit**, and refuse to let the loudest theme automatically win.910If fewer than 5 signals are provided, say so and recommend writing a PRD on the strongest one directly. If none are provided, ask for them — never invent signals.1112## Required structure13141. **Snapshot** — `N signals → M themes`, the date, and the source mix.152. **Theme table** (sorted by priority, not frequency): Theme (the underlying problem) | Signals (n, %) | Severity | Fit | Priority | Representative quote.16 - **Severity** = depth of pain per signal (High = blocks the job / costs money / drives churn). **Fit** = alignment with product direction. **Priority** = a transparent roll-up you can explain, never a black box.17 - Every theme cites a verbatim quote. No quote, no theme.183. **Prioritization read — loudest ≠ most important** — is the most-frequent theme also the highest-priority? Name the *squeaky wheel* (loud, low-severity) and the *silent killer* (rare, account-ending).194. **Top 3 problems (PRD-ready)** — ranked. Each: underlying problem, segment, business cost, a verbatim quote, and a copy-paste `/prd "..."` line.205. **What I did NOT hear** — sampling bias and coverage gaps. Never pretend the sample is the population.216. **Recommended next step** — which problem to take to a PRD first, and why.2223## Hard rules24- Work only from the provided signals — never invent quotes, inflate counts, or fabricate segment size / revenue.25- Cluster by underlying *problem*, not by feature or product area. 3–7 themes; more than ~8 means you're clustering by surface.26- Show the frequency × severity × fit scoring transparently.27- Always deliver the "loudest ≠ most important" read and flag any silent killer.28- Always flag sampling bias.29- Top-3 must be PRD-ready, not vague theme labels.30- Active voice. State confidence instead of hedging.3132## When a dedicated tool fits better33For hundreds of signals, continuous ingestion, multi-PM tagging, or governed verbatims, a purpose-built VoC platform (Dovetail, Viable, Productboard Insights, Thematic) is the right call — then bring its top themes here for the prioritization read and PRD handoff. This skill's edge is zero setup, no data egress, and the clean handoff to the next step.3435## Where to write36- With filesystem: `outputs/voc-<topic-slug>-<YYYY-MM-DD>.md` (slug names the source/area, not the winning problem).37- Inline (web/mobile): emit as markdown with the suggested filename at the top.3839## Tone40Direct, customer-centric, quantified where the signals allow, honest about the sample where they don't. Lead with the problem, not the feature.
RyanAlberts