Claude Skills for Product Discovery: From User Research to Roadmap

Good product discovery has a well-documented shape — plan the research properly, understand what customers are actually trying to accomplish, connect that understanding to a business outcome worth pursuing, and turn the result into a roadmap that can survive contact with a skeptical stakeholder. The frameworks for each stage already exist and are well-established: the Mom Test for interviews, Clayton Christensen's Jobs-to-be-Done, Teresa Torres' Opportunity Solution Tree, outcome-driven roadmapping. The hard part was never knowing the frameworks existed — it was applying them correctly and consistently under a deadline.

These four Claude Skills map directly onto that established methodology, one Skill per stage of the discovery-to-roadmap pipeline, each built on the named framework rather than a generic "do some research" prompt.

Skill 1: Plan Interviews That Produce Usable Insight

Two weeks until a deadline, five interviews scheduled, and about to ask "would you use this if we built it?" — a question that reliably produces enthusiastic nods and zero usable insight, because people are bad at predicting their own future behavior and polite about not wanting to disappoint an interviewer. Good discovery interviews aren't about a clever script; they're about asking only about past, specific behavior, and matching your methodology to what you're actually trying to learn.

The Discovery Interview Prep Skill walks through four adaptive questions — research goal, target segment, constraints, and methodology — and produces a tailored interview plan: opening, core, and closing questions, the specific biases to guard against for that research goal, success criteria for the session, and a recruiting plan. It matches the actual methodology (Mom Test, JTBD-style, switch interviews, journey mapping) to your situation rather than handing you a generic script.

This is a strategic prep process, not a script generator — the output changes meaningfully depending on whether you're investigating churn, validating a new problem hypothesis, or exploring competitive switching.

⏱ Setup takes about 15 minutes. Run it before recruiting starts so the interview guide shapes who you recruit, not the other way around.

Skill 2: Map What Customers Are Actually Hiring Your Product to Do

Ask customers what they want and they'll hand you a feature list — more filters, a dashboard, an export button. Build exactly what they asked for and you often end up with a Slack clone or an email filter that nobody actually loves, because the feature list was never the real need. Jobs-to-be-Done exists precisely to look past the feature request to the underlying job the customer is trying to get done.

The Jobs-to-be-Done Skill structures that exploration across three layers, built on Christensen's JTBD theory and Osterwalder's Value Proposition Canvas: functional jobs (the practical task), social jobs (how the customer wants to be perceived while doing it), and emotional jobs (how they want to feel), alongside a ranked map of pains and gains. The output reframes existing features around real jobs instead of treating each feature request as its own isolated need.

This pairs directly with the Discovery Interview Prep Skill above — interview transcripts are the raw material JTBD analysis works from, and the output here feeds straight into problem statements and epic framing.

⏱ Setup takes about 10 minutes. Best run after a batch of interviews, using the transcripts as source material.

Skill 3: Connect a Business Outcome to a Cheap Experiment

A stakeholder wants feature X, the exec team wants feature Y, and the backlog already has feature Z — and none of them connect to a business outcome anyone can actually move. Teresa Torres' Opportunity Solution Tree forces the conversation upstream: what outcome are we actually driving, which customer problems matter most for that outcome, and what's the cheapest experiment that lets us learn before committing engineering time to building anything.

The Opportunity Solution Tree Skill runs the two-phase process directly: Phase 1 builds the tree itself — one desired outcome, three opportunities beneath it, three candidate solutions under each opportunity. Phase 2 scores each solution on feasibility, impact, and market fit, selects a proof-of-concept, and designs the specific experiment (A/B test, prototype, or concierge MVP) to validate it cheaply.

⏱ Setup takes about 20 minutes. Works best when the JTBD output above feeds directly into the opportunities identified in Phase 1.

Skill 4: Turn Discovery Findings Into a Roadmap That Survives Challenge

A roadmap that's a list of features grouped by quarter is a wishlist with dates, not a roadmap. Engineering builds the features, and the OKRs don't move, because nothing on the list was ever connected to a hypothesis or a measurable outcome — it's just what seemed important when the roadmap was drafted. A real roadmap ties each epic to a hypothesis, an outcome, and a metric, and can explain to an executive why item X was prioritized over item Y.

The Roadmap Planning Skill orchestrates that connection across five phases: gather inputs (OKRs, customer problems — exactly what the discovery Skills above produce), define epics with hypotheses and success metrics attached, prioritize with RICE scoring, sequence into a Now/Next/Later structure with dependencies mapped, and produce the stakeholder-ready communication for presenting it.

⏱ Setup spans 1–2 weeks as a full planning cycle. This is the culmination Skill — it's most powerful when fed directly by the discovery work from the three Skills above.

The Full Discovery-to-Roadmap Pipeline

These four Skills are designed to run in sequence, each one's output becoming the next one's input:

1. Discovery Interview Prep — plan interviews matched to what you actually need to learn
2. Jobs-to-be-Done — turn interview transcripts into functional, social, and emotional job maps
3. Opportunity Solution Tree — connect the jobs and pains identified to a business outcome and a cheap validating experiment
4. Roadmap Planning — turn validated opportunities into an outcome-driven roadmap that can survive a stakeholder challenge

You don't have to run the full pipeline every time — a team mid-cycle might jump straight to the OST if the customer problems are already well understood, or use JTBD on its own to reframe an existing feature backlog. But run end to end, these four Skills replace "we think we should build this" with a defensible chain of evidence from a real customer conversation to a specific line on the roadmap.