WhoCites

WhoCites Methodology — How the 7-Engine AI Visibility Scan Works

WhoCites measures AI visibility across 7 engines using a reproducible, public methodology: prompt discovery, mention detection, rank weighting, share of voice, citation rate, and engine coverage. Each scan returns one 0-100 visibility score, an engine-by-engine breakdown, and a ranked fix list.

Last updated

Run an AI visibility scan

How WhoCites measures AI visibility

WhoCites measures AI visibility by running the same buying-intent prompts across ChatGPT, Claude, Gemini, Grok, Copilot, Perplexity, and Google AI Overviews, then scoring whether each engine mentions the brand, how it ranks it, which competitors it surfaces, and which sources it cites. The output is a single 0-100 visibility score plus the per-engine evidence behind it.

The 7 engines covered

ChatGPT (OpenAI), Claude (Anthropic), Gemini (Google), Grok (xAI), Copilot (Microsoft), Perplexity, and Google AI Overviews. The mix spans browsing-grounded and training-grounded retrieval pipelines so cross-engine variance is visible.

Prompt discovery

Prompts are generated from the brand's category, buyer, and use case (inferred from the public site) and combined with category-level buying-intent templates: comparison prompts, recommendation prompts, and category-fit prompts. The customer does not need to bring a list.

Mention detection

Mention detection identifies the brand by canonical name and known variants (legal name, alternate names) in each engine's response. Detection treats partial mentions and full mentions differently; only meaningful, brand-identifying mentions count toward the score.

Rank weighting

When a response mentions multiple brands, rank weighting assigns more credit to earlier mentions. Being named first in a recommendation response is meaningfully different from being named fifth, and the score reflects that.

Share of voice

Share of voice divides the brand's mention count by total mentions of all brands across the scan. It shows whether the brand dominates the category answer set or is a minority voice next to better-cited competitors.

Citation rate

Citation rate is the percentage of source URLs (where engines expose them) that point to the brand's own domain versus competitor and third-party domains. It measures direct retrieval lift, not just brand mention.

Engine coverage

Engine coverage is the count of engines (out of 7) that mention or cite the brand at all. A brand covered by 6 engines and missing on 1 has a targeted problem; a brand covered by 1 has a systemic problem. Coverage shapes the fix list.

The 0-100 visibility score

The visibility score is a weighted aggregate of per-engine mention rate adjusted for brand rank, normalized to 0-100. It is comparable across scans of the same brand and roughly comparable across brands in the same category. The underlying per-prompt evidence remains in the report so the score is auditable.

Visibility score formula

The score is calculated as weighted brand mentions divided by weighted scored prompts, multiplied by 100. Core prompts carry a weight of 1.0, adjacent prompts carry a weight of 0.35, and off-category prompts carry a weight of 0. This prevents broad or irrelevant prompts from dominating the result.

Engine mention-rate math

Each engine receives its own mention rate using the same prompt-fit weights. If a brand appears in one of four equally weighted core prompts on ChatGPT, the ChatGPT mention rate is 25%. If it appears only in adjacent prompts, that movement is recorded but discounted.

Citation-rate math

Citation rate is calculated only when an engine exposes source URLs. The numerator is source URLs pointing to the scanned brand's own domain; the denominator is all source URLs exposed for the scored responses. Engines that do not expose citations are not treated as zero-citation engines.

Share-of-voice math

Share of voice counts brand and competitor mentions across scored responses, using the same prompt-fit weights as the main score. A competitor mentioned in a core prompt counts more than a competitor mentioned only in an adjacent prompt.

What the scan does NOT measure

WhoCites does not measure code quality, security posture, accessibility compliance, launch readiness, Google Search rankings, paid ad performance, social engagement, or operational metrics. It measures only public AI and search visibility. For pre-launch readiness — robots.txt, sitemap, schema, llms.txt, security headers, and the other machine-readable inputs an AI engine needs before it can parse a product — the complementary pre-launch diagnostic is shippingszn. See shippingszn.com/machine-trust-for-startups for the signals shippingszn checks, shippingszn.com/improve-ai-recommendation-probability for the two-layer model that pairs shippingszn with WhoCites, and shippingszn.com/methodology for the scanner's auditable pipeline. shippingszn inspects the inputs; WhoCites measures the output.