# Customer Discovery

> Design and synthesize customer interviews, problem research, evidence ledgers, and validation experiments. Use when founders need to learn whether a real customer has an urgent problem or test a risky assumption; do not use it to manufacture testimonials, pitch during research, or replace product analytics.

- Skill: `vaibhav0806/customer-discovery` (Agent Skill, multi-file: 5 files)
- Install (CLI): `npx skillmds@latest add vaibhav0806/customer-discovery`
- Raw SKILL.md: https://api.skillmd.com/api/skills/vaibhav0806/customer-discovery/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Product & Planning
- Author: vaibhav0806 (https://skillmd.com/u/vaibhav0806)
- Updated: 2026-09-17
- Page: https://skillmd.com/skills/vaibhav0806/customer-discovery

---


# Customer Discovery

Produce decision-grade evidence, not a list of agreeable quotes.

## Workflow

1. State the decision and the riskiest assumption it depends on.
2. Define the target participant and disqualifiers before recruiting.
3. Read [interviews.md](references/interviews.md) to prepare or conduct interviews.
4. Read [evidence-synthesis.md](references/evidence-synthesis.md) to combine notes without losing contradictions or source context.
5. Read [validation-experiments.md](references/validation-experiments.md) when interviews alone cannot test the assumption strongly enough.
6. Return what is `observed`, `inferred`, `proposed`, `stale`, or `conflicting`; the strongest disconfirming evidence; confidence; and the next decision.

## Boundaries

- Ask about recent behavior, actual workflow, triggers, attempted solutions, cost, and authority. Hypothetical willingness and compliments are weak evidence.
- Do not pitch during a discovery interview or teach the participant the desired answer.
- Route market, ICP, business-model, or resource-allocation choices to `startup-strategy` after evidence exists.
- Route measurement design, funnels, cohorts, and retention proof to `startup-metrics`.
- Minimize personal data. Work in three explicit stages: (1) anonymous criteria and source planning; (2) compiling identifiable prospects; (3) outbound contact. Obtain scoped authorization before stages 2 and 3. Stage 2 authorization must cover target population, permitted source systems, data fields, storage/access, and retention. Stage 3 authorization must cover the exact population, channel, message, incentive, recording, retention, and note sharing. An imperative such as “find and message” does not waive the immediately-before-action check.