Verified, Likely, Unknown: how PMFKit labels every claim

If we don't know, we say we don't know.

Why we label

So you can disagree. And so we don't overstate. When a claim is inferred or we're missing data, the label tells you. You can trust Verified, question Likely, and treat Unknown as a to-do or a gap.

Verified

We observed it. On your site (e.g. a headline, a price, a testimonial) or from a source we name (e.g. search result, traffic estimate). The report can point to the page or the evidence. When we say Verified, we're not guessing.

Likely

We inferred it from evidence. We say why. For example: "Likely B2B given job titles on the site." You can accept it or push back. The label is there so you don't treat it as fact without thinking.

Unknown

We don't have evidence. We might say what would resolve it (e.g. "Add a clear 'Who this is for' section" or "We couldn't find pricing—add a pricing page"). Unknown doesn't mean wrong; it means we're not stating a claim. Use it as a signal for what to fix or what to learn.

How to use it

Trust Verified when you need a solid fact. Question Likely when the inference matters for a decision. Treat Unknown as a list of gaps: either fix the site so we can observe it, or run an experiment to learn. The verdict and one thing this week can still be useful when some claims are Likely or Unknown—we're just being clear about the foundation. Evidence-based vs consultant and how we analyze your site go deeper.