customer-research

Trigger phrases

Phrases that activate this skill when typed to Claude Code:

• customer research
• ICP research
• voice of customer
• review mining
• analyze transcripts

What it does

customer-research is a Claude Code skill from Corey Haines’s marketing-skills repo. It turns Claude into a customer researcher who works two modes: synthesizing existing assets (interview transcripts, support tickets, NPS comments) and gathering new research from public sources (Reddit, G2, Capterra, forums, competitor reviews). The skill activates when you mention “customer research”, “ICP research”, “voice of customer”, or “review mining”.

The output of a session is a research synthesis: jobs-to-be-done statements, pain-point clusters, exact-language quotes you can lift into copy, persona drafts grounded in actual quotes, and a “what surprised us” section that captures the things you didn’t know to ask about.

When to use it

Reach for it when:

• You have 20+ interview transcripts and no time to read them all
• You’re rewriting positioning and need actual customer language, not stakeholder paraphrase
• You’re entering a new segment and want to mine the watering holes (subreddits, communities, review sites)

When not to reach for it:

• You haven’t talked to anyone yet — the skill synthesizes, it doesn’t conjure
• You need quantitative survey analysis at scale — different tooling

Install

The skill is distributed via Corey Haines’s marketing-skills repo. Install via the repo’s recommended path — copy the customer-research SKILL.md into your project’s .claude/skills/customer-research/ directory, or use the repo’s plugin install if you’ve set it up.

Once installed, the skill activates on the trigger phrases above. The first time it runs, it will check for .agents/product-marketing-context.md (or .claude/product-marketing-context.md) — populating that file with your product context first dramatically improves output quality across all of Haines’s marketing skills.

What a session looks like

A typical session has three phases:

1. Source intake. Transcripts, tickets, NPS comments, review URLs, subreddits. The skill catalogues what you’ve got and asks what’s missing.
2. Pattern extraction. JTBD statements, pain clusters, language patterns, objections. Every claim is anchored to a quote — no paraphrase masquerading as insight.
3. Synthesis output. Persona drafts with quotes, a copy-ready language doc (exact phrases customers used), and a research gaps list (what you’d need to talk to next).

The discipline that makes it work: every insight has to be quote-anchored. The skill refuses to assert “customers want X” without showing the line where they said it.

Receipts

Honest reporting on what customer-research produces and where it has limits:

Where it works well:

• The quote-anchored discipline is the difference between research and confirmation bias
• Watering-hole mining (Reddit, G2) surfaces objections you’d never get in a sales-friendly interview
• Persona drafts grounded in quotes are dramatically more useful than HR-poster personas

Where it backfires:

• Public review sites have selection bias — only the very-happy and very-angry post
• Without a meaningful sample (say, fewer than 5 transcripts) the patterns are anecdotes, not signal

Pattern that works: before any major copy or positioning project, run a one-hour research synthesis on whatever you already have. The exact-language doc alone justifies the session.

Source and attribution

Originally written by Corey Haines. The canonical SKILL.md and any supporting files live in the customer-research folder of his marketing-skills repository.

License: MIT. You can install, adapt, and redistribute the skill, with attribution preserved.

This page documents the skill from a practitioner’s perspective. For the formal spec and any updates, defer to the source repo.