What each section covers
A greenfield validation runs 30 sections; an upgrade analysis runs 5. Each section has a single, narrow question it answers, a model picked for that kind of work, and a credit cost tied to reasoning intensity. The table below mirrors the order shown in the report, grouped by research phase.
Phase 1 — Problem & Customer
Who hurts, how badly, and whether they will pay to make it stop.
Is the pain real, frequent, and expensive enough to act on?
Who exactly buys this? Role, budget authority, day-in-the-life.
What progress is the customer trying to make? Ulwick 4-force / switch interview frame.
Honest scorecard on why you specifically are credible here.
Mom-Test questions tailored to your personas, ready-to-use script.
Phase 2 — Market
How big the pie is and whether the timing is right.
Top-down and bottom-up market sizing with named sources.
Tailwinds, unlocks, regulatory shifts, behaviour changes that make today different.
Search trends, forum complaints, growing competitor traffic — proof of pull.
Adjacent winners and dead-twin failures — what they teach.
Phase 3 — Competition
Who is already in the room and where you can outflank them.
Up to 8 competitors with opportunity score, position, key clusters, whitespace wedge.
What makes customers stick to incumbents — and how you make leaving cheap.
Domain availability via RDAP, social handle availability, trademark gut-check.
Phase 4 — Economics
Whether the unit math works at the price your buyer accepts.
Van Westendorp price-sensitivity meter with optimal/acceptable bands.
CAC, LTV, payback, gross margin — formula trail shown, not just the answer.
Ranked acquisition channels with CAC estimate and saturation risk per channel.
Keyword opportunity surfaces, content gaps, who owns the rankings today.
Phase 5 — Risk, Legal & Action
What can kill the idea before launch, and the first experiments to run.
Industry-specific permissions, data-handling rules, jurisdictional disclaimers.
Entity choice, co-founder vesting basics, IP assignment, tax notes per jurisdiction.
Where the chosen stack constrains future moves — vendor, license, migration cost.
Hiring depth in your geography, comp ranges, build-vs-contract trade-offs.
8-12 risks scored by likelihood × impact with named mitigations.
First 3-5 cheap experiments with hypothesis, method, and pass-fail criteria.
Synthesis layer
Strategic frames that turn raw findings into decisions.
Four-quadrant SWOT plus TOWS strategy crosses (SO/ST/WO/WT).
Each force scored 1-10 with founder lever to soften it.
9-element Osterwalder canvas plus the single riskiest assumption to test first.
0-100 composite across 5 sub-dimensions with raisers per dimension.
Market type, wedge, buyer, direction, 5 strategic recos — the digest.
Cross-cutting outputs
The verdict and the talk-track that wrap the report.
BUILD_NOW / BUILD_WITH_PIVOT / RECONSIDER / DO_NOT_BUILD with score and one-paragraph blurb.
Brutal-but-kind 4-paragraph callout of the riskiest assumptions in your pitch.
Investor-friendly 4-paragraph narrative pulling everything together.
Upgrade-kind sections (existing product analysis)
Used when you connect an existing repo or live URL to plan v2/v3 instead of starting fresh.
Stack, dependencies, traffic signals, infra footprint, observed feature set.
Where the existing product sits in its category and how the market reads it.
Architectural risks, modernisation targets, and lock-in to address before v2.
Where the next version should invest — keep/deepen/drop/rebuild classification per surface.
MAJOR_UPGRADE / INCREMENTAL / PIVOT / MAINTAIN with score and reasoning.
Pricing breakdown: see credits & per-section pricing for full totals. Scoring: how the verdict is computed.