How PMFKit analyzes your site (no black box)

No magic. We look at your site and a few external signals. Then we label what we're sure about and what we're not.

Discovery

We find pages. Sitemap first, then conventional paths: /pricing, /about, /features, /blog. We cap how many we crawl so the run finishes in a few minutes. What we collect: headlines, key points, pricing tables, team info, customer logos, metadata.

Enrichment

We add external data when we can: brand search (do you rank for your name?), competitors, traffic estimate, value-prop clarity. When an API fails or we hit a limit, we fall back. We never block the analysis; we just mark what we have and what we don't.

Five components

Brand (confusability, searchability, recall—all in terms of distribution impact). Product (archetype, value prop, target user, depth). Pricing (readiness, visibility, fit). Market (category, positioning, competitive context). Distribution (channels we see, pattern fits, DRL). Each component produces structured output and, where it makes sense, claims with Verified / Likely / Unknown.

Synthesis

One step combines the five into a verdict: stage, Continue / Pause / Kill, reasoning, focus area, next milestone, one thing this week, blockers, advantages. We use the same vocabulary in the report so the narrative matches the engine.

We show our work

The report links conclusions to evidence. You see which pages we used, which claims are Verified vs Likely vs Unknown, and what we inferred. So you can disagree. Verified, Likely, Unknown and how to read your report go deeper.