The Continue, Pause, Kill verdict
One of three calls, with a confidence label, grounded in the evidence chain.
Most PMF tools equivocate. Generic LLMs are trained to hedge. Hourly-billed advisors have a weak incentive to deliver a Kill (the recommendation that ends the engagement contradicts the business model that pays for it). PMFKit is built around the verdict because nobody else is structurally positioned to deliver it honestly.
Continue
The evidence supports the current path. There's a real product, real signal, and a clear next focus area. Continue does not mean "you have PMF." It means "keep investing on the current vector and the named focus is the right one for the next 30 to 90 days."
What to do. Read Strategies for the focus area and milestones. Pick one milestone you can validate or kill in two weeks. Re run PMFKit at your next decision moment (board update, pre fundraise, pivot trigger) to see if the signal is still pointing the same way.
Pause
The evidence is mixed. Something specific needs to be tested or fixed before more capital, hiring, or runway goes into the current direction. Pause is the most common verdict for founders one to three months past a pivot or one month before a board update.
What to do. Strategies will name the one thing that needs to be true for Continue. Don't fix everything; fix the one thing. If the named test isn't possible (no users to survey, no traffic to A/B), Strategies will recommend the cheapest path to the smallest version of the test. Re run PMFKit after the test resolves.
Kill
The evidence says this isn't working. The current positioning, market fit, or distribution channel is materially misaligned with the product, and the cleanest move is to stop investing on this vector. Kill is rare and structurally weighted: PMFKit will say so explicitly when it does, and the confidence label tells you how strong the underlying evidence is.
What to do. Read the Strategies report carefully. A Kill is not a recommendation to fold the company; it's a recommendation to stop investing on this specific vector. Strategies will name the cleanest pivot path or the cleanest exit. If the verdict feels wrong, scroll to Discoveries and challenge a specific claim. The trust pill on each claim tells you how confident the evidence is.
The confidence label
Every verdict carries one of three confidence labels:
- High confidence. The verdict rests on claims that are mostly Verified.
- Moderate confidence. The verdict mixes Verified and Likely claims, with limited Unknown.
- Low confidence. The verdict rests on a majority of Unknown claims. PMFKit will explicitly recommend re running after you provide additional input or after time has elapsed and more public signal is available.
See trust pills for the per claim labels that feed into the verdict confidence.
When to disagree
You may know things PMFKit cannot see. The advisory disclaimer is in every report's footer for a reason: the verdict reflects the model's synthesis of public footprint data at a point in time. The point of the trust pills and the full evidence chain is that you can disagree with a specific claim, not just the conclusion.