Developing Product Intelligence
Markdown--- name: developing-product-intelligence description: Transforms raw product ideas into comprehensive, research-backed product concepts including problem validation, market research, user research, PRDs, user stories, MVP scope, and roadmaps. Use when a user proposes a new app/game/website idea, asks for a PRD, requests product strategy, competitor analysis, user personas, feature prioritization, or MVP definition. --- # Master Product Intelligence Combines Product Discovery, PRD Generation, Market & Trend Research, User/UX Research, and Product Strategy & MVP Planning into one end-to-end workflow that turns a raw idea into a build-ready product concept.
Use this skill when the user:
- Pitches a raw app/game/website idea ("I want to build...")
- Asks for a PRD, product spec, or requirements document
- Requests competitor/market analysis for a product
- Asks for user personas, user journeys, or UX research
- Needs help defining MVP scope or feature prioritization
- Asks for a product roadmap or risk analysis
- Says things like "validate my idea," "is this a good product," "help me plan this product"
Do NOT use this skill for: pure coding tasks, bug fixes, or UI implementation unrelated to product definition — redirect to normal coding workflow instead.
- Never fabricate data. No invented market sizes, statistics, user counts, or competitor numbers. Every claim must be tagged:
[FACT - source: ...](only if real research/browsing available)[ASSUMPTION](explicitly reasoned guess, must be validated later)[RECOMMENDATION](your expert suggestion, not a fact)
- Ask before assuming. If critical info is missing (see Required Inputs below), ask targeted questions — batched together, asked ONCE. Never re-ask something already answered.
- No generic templates. Every output must reference the specific idea, specific users, specific features. Reject boilerplate phrasing like "users want a seamless experience."
- Research if possible. If web access is available, search for current competitors, trends, and benchmarks before writing Market Research or Competitor Analysis sections. If not available, clearly mark the section as "ASSUMPTION-based — recommend validating via real research."
- Sequential gates. Do not jump straight to a full PRD before Discovery is validated. Follow the workflow order unless the user explicitly asks to skip ahead (then warn about the risk of skipping validation).
Before starting, ensure you know:
- The core idea — what is the product, in the user's own words?
- The problem it solves — if unclear, ask "What problem does this solve, and for whom?"
- Target platform — app / game / website / all?
- Target audience (if known) — age, role, context of use
- Business model intent (if known) — free, paid, ads, B2B, B2C?
- Stage — brand new idea, or existing product needing a feature/pivot?
- Scope/time constraint (if known) — solo dev, small team, deadline?
If 3+ of these are unknown, ask a single consolidated question block before proceeding. Never interrogate one question at a time.
Progress:
- [ ] Step 0: Gather required inputs (ask if missing)
- [ ] Step 1: Product Discovery (problem, audience, personas)
- [ ] Step 2: Market & Trend Research (competitors, trends, gaps)
- [ ] Step 3: UVP & Differentiation
- [ ] Step 4: PRD Draft (overview, requirements, user stories, AC)
- [ ] Step 5: Feature Prioritization & MVP Scope
- [ ] Step 6: Roadmap
- [ ] Step 7: Risk & Assumption Analysis
- [ ] Step 8: Quality Checklist review before delivering final doc
Step 0 — Gather Inputs
Ask the consolidated question block if needed. Proceed with explicit [ASSUMPTION] tags for anything the user says "not sure" to, and note it must be validated.
Step 1 — Product Discovery
- Restate the idea in one clear sentence.
- Write the Problem Statement: who has the problem, what is the problem, why existing solutions fall short, what happens if unsolved.
- Define Target Audience (primary + secondary segments, with reasoning).
- Build 1-3 User Personas (name, role/context, goals, frustrations, a representative quote, tech comfort level). Tie each persona back to the problem statement — no generic personas.
Step 2 — Market & Trend Research
- If research tools available: search for 3-5 real competitors (direct + indirect) and current relevant trends in this product category.
- For each competitor capture: what they do well, where they fall short, pricing model (if public), and the gap relative to the user's idea.
- Summarize 2-3 relevant technology/design/industry trends affecting this product (e.g., AI features, platform shifts, regulation) — tag as FACT with source or ASSUMPTION if unverified.
- Identify explicit market gaps/opportunities the idea could exploit.
Step 3 — Unique Value Proposition
- Write a one-sentence UVP: "For [target user], who [problem], [product] is a [category] that [key benefit], unlike [main alternative] which [weakness]."
- List 3-5 concrete differentiators grounded in Step 2's competitor gaps.
Step 4 — PRD Draft
Use the PRD Template below. Populate every section with specifics from Steps 1-3. No placeholder text.
Step 5 — Feature Prioritization & MVP Scope
- List all candidate features from discovery/PRD.
- Score each using RICE (Reach, Impact, Confidence, Effort) or MoSCoW (Must/Should/Could/Won't) — pick MoSCoW for early-stage ideas, RICE if the user has usage/data signals.
- Define MVP Scope: the minimum feature set that lets the core problem be solved and validated. Explicitly list what is DEFERRED and why.
- Define Phase 2+ features separately.
Step 6 — Roadmap
- Present a phased roadmap: MVP → V1 → V2 (or Now/Next/Later if timeline is unknown).
- Each phase: goal, key features, success signal that triggers moving to next phase.
Step 7 — Risk & Assumption Analysis
- List unproven assumptions (market demand, willingness to pay, technical feasibility, behavior change required).
- List risks: market risk, technical risk, business/monetization risk, UX adoption risk.
- For each: impact (H/M/L), likelihood (H/M/L), and a mitigation or validation method (e.g., "test with 10 user interviews before building").
Step 8 — Quality Checklist
Run the Quality Checklist (below) before presenting final output. Fix gaps before delivering.
Markdown# Product Requirements Document: [Product Name]
[2-4 sentences: what it is, for whom, core value]
[Specific problem, who experiences it, current alternatives and their shortcomings]
Primary: [segment + rationale] Secondary: [segment + rationale]
Persona 1: [Name]
- Context/Role:
- Goals:
- Frustrations:
- Quote: "..."
| Competitor | Strengths | Weaknesses | Gap We Exploit |
|---|
[One-sentence UVP + differentiators list]
- FR1: [System shall...]
- FR2: ...
- Performance:
- Security:
- Scalability:
- Usability/Accessibility:
- Platform/Compatibility:
US1: As a [persona], I want to [action], so that [benefit].
Acceptance Criteria:
- Given [context], when [action], then [result]
- ...
| Feature | Priority (MoSCoW/RICE) | Rationale |
|---|
Included: [...] Explicitly Deferred: [... with reason]
- MVP: [goal, features, success metric]
- V1: [...]
- V2: [...]
| Risk/Assumption | Type | Impact | Likelihood | Mitigation |
|---|
[List anything tagged ASSUMPTION that needs real-world testing]
Input: "I want to build an app that helps freelance designers manage client feedback without messy email threads."
Output (abridged):
- Problem Statement: Freelance designers lose time and context switching between email, WhatsApp, and files to collect client feedback on design revisions; feedback gets lost or misattributed to wrong file versions.
- Target Audience: Primary — solo freelance designers (1-5 active clients); Secondary — small design agencies (2-10 staff).
- Persona: "Rani, 29, freelance brand designer, juggles 4 clients, loses 3-5 hrs/week consolidating feedback from Gmail + WhatsApp screenshots."
- Competitor gap: Tools like Figma comments require client to have Figma literacy; tools like ReviewBoard are agency-priced, too expensive for solo freelancers → Opportunity: lightweight, email-free, link-based feedback tool priced for solo freelancers.
- UVP: "For solo freelance designers who lose feedback in scattered channels, FeedbackFlow is a link-based review tool that centralizes all client comments by version, unlike Figma/agency tools which require logins or enterprise pricing."
- MVP Scope: Upload design version → shareable client link → click-to-comment on image → version history. Deferred: real-time chat, team roles, billing integration.
- Risk:
[ASSUMPTION]clients will adopt a new tool instead of replying by email — Impact: High, Likelihood: Medium — Mitigation: test with 5 real freelancer-client pairs before full build.
Before delivering final output, verify:
- Problem statement names a specific user and specific pain, not generic
- Personas are differentiated and tied to the problem (not interchangeable)
- Competitor analysis has real or clearly-flagged-assumption data, not invented stats
- UVP is one sentence, specific, and contrasts against a named alternative
- Every functional requirement is testable/verifiable
- Every user story has at least one Given/When/Then acceptance criterion
- MVP scope excludes at least one tempting feature, with stated reason
- Roadmap phases have a measurable trigger to advance, not just a feature list
- All risks have impact + likelihood + mitigation, not just a description
- No section uses filler/generic language that could apply to any product
- All unverified claims are explicitly tagged ASSUMPTION or RECOMMENDATION
- If idea is too vague (e.g., "I want to build an app"): ask the consolidated input questions before any output.
- If user asks only for one piece (e.g., "just give me user personas"): deliver only that section, but still ground it in Step 1 discovery logic — don't skip reasoning, just skip unrequested sections.
- If data/research is unavailable: never invent numbers. Say "research tools unavailable — this is a reasoned AS