# Product Customer Discovery

> Plan and run product-focused customer discovery interviews and synthesize actionable customer insights. Use when you need to define discovery goals/hypotheses, identify an ICP/segments, create a recruiting plan, write an interview guide, conduct interviews, analyze notes (themes, JTBD, pains/gains), and produce a discovery readout with opportunities, risks, and next experiments.

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

---


# Product Customer Discovery

## Purpose
- Reduce product/market risk by learning how target users behave today, what problems they truly have, and what they already do to solve them.
- Turn qualitative conversations into decisions: who to build for (ICP), what to solve (problem framing), and what to test next (experiments).

## Quick triggers
Use this skill when the user asks for:
- “product customer discovery”, “user research”, “problem interviews”, “exploratory interviews”
- “write an interview script/guide”, “what questions should I ask users”
- “define ICP/personas/segments”, “JTBD”, “pain points”, “opportunity sizing (qual)”
- “synthesize interview notes”, “extract themes/insights”, “create a discovery report”

## Inputs to ask for (minimum)
1. Product/service and stage (idea, MVP, growth) + decision(s) discovery must unblock
2. Target audience hypotheses (who) and problem hypotheses (what/why)
3. Constraints: timeline, number of interviews, geography/language, incentives, recruiting channels

## Outputs (suggested)
- **Discovery plan**: goals, hypotheses, target segments, recruiting criteria, timeline
- **Interview guide**: opening, context questions, story prompts, probing, wrap-up
- **Synthesis**: themes + evidence (quotes), JTBD/pains/gains, opportunity areas, risks/unknowns
- **Next steps**: prioritized experiments (e.g., landing page, concierge test, prototype test)

## Core workflow (end-to-end)
1. **Align on outcomes**: confirm what decision will be made from the research and what “good evidence” looks like.
2. **Define hypotheses**: write 5–10 falsifiable statements (ICP, problem, willingness, constraints, alternatives).
3. **Select participants**: define inclusion/exclusion criteria, quotas across segments, and screening questions.
4. **Design the interview**:
   - prefer “tell me about the last time…” over “would you use…”
   - focus on current behavior, existing alternatives, constraints, and consequences
5. **Run interviews**:
   - start with rapport + consent; keep it conversational
   - ask for specific incidents; probe for frequency, severity, triggers, and workarounds
   - capture verbatims and observable facts; separate facts from interpretations
6. **Synthesize**:
   - affinity-map notes into themes; label with evidence + confidence
   - map to JTBD (situation → motivation → desired outcome) and pains/gains
   - identify “strong signals” (repeated patterns, costly workarounds, high stakes)
7. **Decide & recommend**:
   - rank opportunities by severity, frequency, reachable audience, and differentiation
   - propose next experiments with clear success metrics and cheapest test first
8. **Share readout**: present insights, what changed vs. assumptions, open questions, and the plan.

## Interview guide template (outline)
1. **Intro**: who you are, purpose, confidentiality, recording consent, timebox
2. **Background**: role, context, responsibilities, tools/workflow
3. **Story prompts** (core): “Walk me through the last time you…”
4. **Probing**:
   - triggers: “what started this?”
   - frequency: “how often?”
   - severity: “what happens if you don’t solve it?”
   - alternatives: “what did you try? why that? what did it cost?”
   - decision-making: “who’s involved? what’s the budget/approval path?”
5. **Wrap-up**: biggest pain, ideal outcome, who else to talk to, follow-up permission

## Quality checklist
- Goals and hypotheses are explicit and falsifiable
- Participant criteria and screening reduce bias (no “friends and fans” only)
- Questions avoid leading language and future hypotheticals
- Notes separate verbatims/facts from interpretations
- Insights are backed by evidence, not anecdotes
- Recommendations include next experiments and success metrics

## Common mistakes (avoid)
- Asking for feature opinions instead of behavior (“Would you use X?”)
- Interviewing only easy-to-reach users and generalizing
- Treating one loud quote as a “theme” without triangulation
- Skipping the decision step (insights without a recommendation and next tests)