Methodology overview
The AI Visibility Score is derived from a fixed test pipeline against four model providers. The full methodology is published; this page is the executive summary.
Query four model providers
The system queries ChatGPT, Claude, Gemini, and Perplexity with a fixed prompt set drawn from buyer-intent questions in the brand's category. Each prompt is repeated five times per provider to account for non-deterministic output.
Run five analysis passes per response
Pass 1 matches brand names with normalized casing and TLD handling. Pass 2 records list position. Pass 3 classifies per-brand sentiment per response. Pass 4 applies a deterministic point rubric. Pass 5 measures citation reach against the target domain.
Score per provider, aggregate uniformly
Each provider produces an independent 0–100 score with a sample size. The headline number is the uniform-weight aggregate across providers. A separate citation score reports target-domain reach across the run.
Publish the report with full traceability
Results include per-provider breakdowns, the full prompt set used, the pinned model versions queried, and the deterministic rubric applied. Every score is reproducible from the bytes documented in the methodology.
The methodology page lists the prompts, pinned model versions, rubric, and changelog. Score comparability across runs is preserved by version-pinning the engine and the model registry.