How we arrive at confidence.

Signals

Three classes of signals. DOM structure — the layout and copy your page renders. Public positioning — your landing, your docs, your nearest competitors. API traces (with the Chrome Extension) — what your app actually returns when someone clicks.

A Deep Report typically surfaces dozens of signals pulled from across your key pages.

Angles

Each angle is a messaging frame — Pain → Solve, Before → After, Risk → Guarantee, Benchmark → Outcome. We score each candidate on three axes:

• Category relevance — does it fit your product's job-to-be-done?
• Competitive distinctiveness — does it separate you from the nearest alternatives we scanned?
• Proven-frame match — does it match patterns that have worked in B2B SaaS short-form?

Top 3 angles surface in every report, with the strongest highlighted as the best match.

Confidence

We don't publish the exact weights or the scoring formula — they'd become an attack surface for SEO-style gaming and erode the score's value as a quality signal. Here's what we do publish:

Confidence blends three inputs:

• Signal coverage — how much of your product we could actually read.
• Cross-source alignment — how consistent the evidence is across pages.
• Internal evaluator pass — an automated QA step before the report surfaces.

Every report is gated on a confidence threshold. Weaker findings are moved to an exploratory section rather than surfaced as primary claims. You'll see the confidence score, not the threshold — we keep that private so it stays a quality floor, not a competitive benchmark.

Cited sources

Every signal carries its source page so claims stay traceable. Free Surface Reports use the same engine on a smaller scope.