Guide
How to check product-market fit with ChatGPT or Claude
ChatGPT or Claude can check where your product stands on the way to product-market fit, as long as it grades your evidence rather than your hopes. Dan Olsen's pyramid gives it five layers to work up through, each resting on the one below: target customer, underserved needs, value proposition, feature set and user experience. Ask for a hypothesis and an evidence grade at each layer (strong for what customers do, weak for opinions or a few anecdotes, none), then the lowest weak layer and one test for it, with the result that would confirm or change the hypothesis. A polished user experience can't rescue a vague customer, so the test goes where the pyramid is weakest.
Short answer
- Five layers, from the customer up.
- Grade the evidence: behavior beats opinion.
- Fix the lowest weak layer first.
- One test, with what would change your mind.
The brief to copy
Product-market fit review brief
"Review our path to product-market fit with Dan Olsen's five layers: target customer, underserved needs, value proposition, feature set, UX.
For each layer, write our hypothesis, the evidence for it from what I paste, and a grade: strong (customers' behavior), weak (opinions or a few anecdotes) or none. Write [no evidence] where there's nothing; never invent an interview result or a metric.
Keep needs apart from features. Map each feature to a need, and list the features that serve no chosen need and the needs with no feature. Name the lowest layer graded weak or none, and plan one test for it: who, how many, what result confirms the hypothesis and what result changes it.
Our product, users, interviews and numbers: [paste]"
A worked example
Illustration: a made-up business. A booking app for dog groomers has 120 paying accounts after eight months: 70 mobile groomers and 50 salons. A third cancel within three months. The founders describe the customer as "pet businesses". Its features are online booking, reminders, deposits, stock tracking and loyalty points. Five of seven interviewees named no-shows as their biggest cost. Reminders are used by 90% of active accounts, stock tracking by 6%. The review, in part:
| Layer | Hypothesis and evidence | Grade |
|---|---|---|
| Target customer | "Pet businesses"; 70 mobile groomers and 50 salons pay, but nobody knows which group stays | weak |
| Underserved needs | Fewer no-shows; 5 of 7 interviewees said so | weak |
| Value proposition | [check: not stated] | none |
| Feature set | Reminders used by 90%, stock tracking by 6%; loyalty points [check: usage] | weak |
| UX | [no evidence] | none |
Lowest weak layer: the target customer. "Pet businesses" covers two groups, mobile groomers and salons, whose working days and needs may differ. Next test: split three-month retention between mobile groomers and salons, and interview eight accounts that stayed past three months. It confirms the hypothesis if one group stays clearly more and names the same need. It changes the hypothesis if both stay at the same rate, which would send the founders back to the needs layer. Stock tracking and loyalty points go on the cut list until then.
What stays with you
- Bring the evidence: interview notes, usage by feature, cancellations. The AI grades only what you paste.
- Leave gaps empty; "[no evidence]" is a finding.
- Talk to the customers yourself; the test is yours to run.
- Refer to accounts by ID, not names, when you paste usage data.
- Run the review again after each test, from the bottom layer.
Next step: a project that keeps the evidence
A fit review gets better each time you add a test's results. A Claude project for founders keeps your interviews, numbers and past reviews in one place, so the next review starts from them.
Skills for this
- ProductProduct-market fit pyramid reviewWrites the hypothesis for each layer of Dan Olsen's product-market fit pyramid, grades the evidence, finds the lowest weak layer, and plans the next MVP test.
- ProductUser interview synthesisSynthesizes interview notes into themes with participant counts, quotes, jobs-to-be-done, insights vs. assumptions, and recommendations with confidence.
- FinanceUnit economics checkChecks CAC, LTV, LTV to CAC, payback, and gross margin from your numbers, shows each formula, and flags definitional mistakes that inflate results.
- ProductA/B test planWrites an A/B test brief: hypothesis, primary and guardrail metrics, MDE, sample size and duration with the math, stop rules, and a decision rule set in advance.
Questions
Can AI tell me whether I have product-market fit?
Not by itself. It can sort your evidence by layer and say where it's thin, but only customers' behavior settles the question. One common measure is Sean Ellis's survey: ask active users how they'd feel if they could no longer use the product. In First Round Review's account, startups where 40% or more said "very disappointed" were the ones that grew.
What is the product-market fit pyramid?
Dan Olsen's model from The Lean Product Playbook (Wiley, 2015). Five layers, each resting on the one below: the target customer and their underserved needs make up the market, and the value proposition, feature set and user experience make up the product. A problem in a low layer can't be fixed by polishing a higher one.
Where do I start if my product isn't catching on?
At the lowest layer whose evidence is weak. Most often that's the target customer: a description too broad to share needs, with nobody sure who the users who stayed are. The test for that layer is usually interviews with the users who stayed, plus their usage, not a new feature.
AI skill
Have your AI find the layer that blocks product-market fit
Put this skill in AI Skill Pod and everyone on your team gets it in their Claude, Cursor or ChatGPT (Pro, Business, Enterprise): one invite link, free to start. Product-market fit pyramid review has your AI review your path to product-market fit the way Dan Olsen's pyramid lays it out: a hypothesis and an evidence grade for the target customer, underserved needs, value proposition, feature set and UX, the lowest weak layer, and one test for it with the result that would change your mind.
14 days free · No card needed
Sources
- Dan Olsen: The playbook for achieving product-market fit (Mind the Product) — the pyramid's five layers: target customer at the base, then underserved needs (the market), then value proposition, feature set and UX (the product) (read 2026-11-05)
- The Lean Product Playbook (book site) — the Lean Product Process: target customers, underserved needs, product strategy, MVP, prototype, test with customers (read 2026-11-05)
- First Round Review: How Superhuman built an engine to find product/market fit — Sean Ellis's question, "How would you feel if you could no longer use the product?"; across nearly a hundred startups, 40% answering "very disappointed" separated those that grew (read 2026-11-05)