customer-research
Mines reviews, interviews, support tickets and forum threads for voice-of-customer insight — verbatim pains, desires, objections and triggers, clustered into themes with JTBD statements and a message-market map. Use when the user says "analyze these reviews", "what do customers a
Install
npx skills add https://github.com/alebgl77/claude-inc/tree/main/skills/customer-research
claude plugin marketplace add https://llmmart.ai/marketplace.json && claude plugin install alebgl77-claude-inc@llmmart
git clone https://github.com/alebgl77/claude-inc.git
The skills CLI installs just this skill, for any of its supported agents. Claude Code installs the whole alebgl77/claude-inc collection as a plugin from our marketplace. Git is the plain clone.
Skill manifest
Customer Research — Voice of Customer
"Synthesise user voice"
When to use
- "Here are 200 reviews — what do customers actually care about?"
- "Synthesize these interview notes / support tickets"
- "What objections keep coming up?"
- Before positioning, landing copy, or ads — copy written without voice-of-customer is guessing
- Works on pasted text or files; optional upgrade: web search to mine public reviews and Reddit threads
Workflow
- Ingest the corpus: pasted text, CSV/exports, or files. Note the source mix and any sampling bias (e.g. only angry customers write tickets).
- Extract verbatims into four buckets: pains, desires, objections, buying triggers. Keep the customer's exact words — never paraphrase at this stage.
- Cluster into themes per bucket; count frequency so loud-but-rare doesn't beat quiet-but-common.
- Write JTBD statements for the top clusters: "When , I want to , so I can ."
- Build the message-market map: their words → your copy blocks (headline candidates, bullet candidates, objection-handling lines) — quoted or lightly compressed, never marketing-speak.
- Flag the gaps: pains competitors' messaging ignores, plus anything surprising that contradicts current positioning.
- Hand off: name the top theme a copywriter should lead with, and the one objection every asset must answer.
Output format
## Voice of Customer — {corpus, n items, sources}
### Themes by frequency
| Bucket | Theme | Freq | Best verbatim |
|--------|-------|------|---------------|
### JTBD (top 3)
1. When ..., I want to ..., so I can ...
### Message-market map
| They say (verbatim) | Use it as |
|---------------------|-----------|
| "..." | Headline / bullet / objection-handler |
### Gaps & surprises
- ...
### Handoff
Lead with: {theme}. Must answer: {objection}.
Quality bar
- Every theme backed by ≥ 2 verbatims, quoted exactly
- Frequencies counted, not vibed
- Sampling bias of the corpus stated up front
- JTBD statements contain a real situation, not a demographic
- Map entries are usable copy blocks, not categories
- At least one finding that challenges the current positioning (or explicit "none found")
Example
Ask: "Analyze these 80 G2 reviews of my scheduling tool." Produced: theme table (top pain: "double-booked because integrations lag", 19 mentions), 3 JTBD statements, a message-market map with 8 ready-to-use lines, gap ("nobody markets to the assistant persona who actually configures it"), and the handoff note.
Files (claude-inc)
-
SKILL.md 3 KB
--- name: customer-research description: Mines reviews, interviews, support tickets and forum threads for voice-of-customer insight — verbatim pains, desires, objections and triggers, clustered into themes with JTBD statements and a message-market map. Use when the user says "analyze these reviews", "what do customers actually want", "synthesize this feedback", or before writing any positioning or copy. --- # Customer Research — Voice of Customer > "Synthesise user voice" ## When to use - "Here are 200 reviews — what do customers actually care about?" - "Synthesize these interview notes / support tickets" - "What objections keep coming up?" - Before positioning, landing copy, or ads — copy written without voice-of-customer is guessing - Works on pasted text or files; optional upgrade: web search to mine public reviews and Reddit threads ## Workflow 1. **Ingest the corpus**: pasted text, CSV/exports, or files. Note the source mix and any sampling bias (e.g. only angry customers write tickets). 2. **Extract verbatims** into four buckets: pains, desires, objections, buying triggers. Keep the customer's exact words — never paraphrase at this stage. 3. **Cluster into themes** per bucket; count frequency so loud-but-rare doesn't beat quiet-but-common. 4. **Write JTBD statements** for the top clusters: "When {situation}, I want to {motivation}, so I can {outcome}." 5. **Build the message-market map**: their words → your copy blocks (headline candidates, bullet candidates, objection-handling lines) — quoted or lightly compressed, never marketing-speak. 6. **Flag the gaps**: pains competitors' messaging ignores, plus anything surprising that contradicts current positioning. 7. **Hand off**: name the top theme a copywriter should lead with, and the one objection every asset must answer. ## Output format ``` ## Voice of Customer — {corpus, n items, sources} ### Themes by frequency | Bucket | Theme | Freq | Best verbatim | |--------|-------|------|---------------| ### JTBD (top 3) 1. When ..., I want to ..., so I can ... ### Message-market map | They say (verbatim) | Use it as | |---------------------|-----------| | "..." | Headline / bullet / objection-handler | ### Gaps & surprises - ... ### Handoff Lead with: {theme}. Must answer: {objection}. ``` ## Quality bar - [ ] Every theme backed by ≥ 2 verbatims, quoted exactly - [ ] Frequencies counted, not vibed - [ ] Sampling bias of the corpus stated up front - [ ] JTBD statements contain a real situation, not a demographic - [ ] Map entries are usable copy blocks, not categories - [ ] At least one finding that challenges the current positioning (or explicit "none found") ## Example **Ask**: "Analyze these 80 G2 reviews of my scheduling tool." **Produced**: theme table (top pain: "double-booked because integrations lag", 19 mentions), 3 JTBD statements, a message-market map with 8 ready-to-use lines, gap ("nobody markets to the assistant persona who actually configures it"), and the handoff note.
Comments (0)
Sign in to join the conversation.
Reviews (0)
No reviews yet.
No comments yet.