MentionAudit Index · Methodology · v1.0.0
How the MentionAudit Index works
The Index measures brand visibility share in AI search engines. It is not a ranking of model quality. The numbers on every leaderboard page describe how often a brand surfaces in answers to category-defining prompts during a sampling window — and nothing more.
What the Index measures
For each vertical (Scheduling, AI coding tools, CRM, and so on), the Index runs a fixed prompt set against multiple AI search engines and parses each response for two things:
- Mention rate— the percentage of the vertical's prompts in which a brand was named.
- Citation rate— the percentage in which a brand's own domain was linked.
Public leaderboards show aggregated counts only. We never republish raw model outputs on public pages — verbatim responses live only inside authenticated customer dashboards.
Engines covered
The public Index v1 reports rankings sourced from ChatGPT, Claude, and Gemini. Perplexity results are available inside authenticated customer dashboards but not on public leaderboard pages, in keeping with Perplexity's API terms. Copilot is not included in the public Index v1; we may add it back via Azure OpenAI in a future methodology version.
Sampling cadence
Every prompt × engine pair runs daily. Reported rankings are 7-day rolling averages. We publish the exact sampling window (date-from / date-to) on every leaderboard page so any reader can reproduce the test by running the same prompts on the same dates.
Brand & prompt selection
Brands per vertical are selected from category leaders on G2 and Capterra, brands meeting a Crunchbase ≥$5M funding signal, and brands we observe co-occurring in AI-search answers for the category's flagship prompts. Each list also reserves slots for emerging entrants to keep the Index reflective of the current market rather than the 2010-era incumbents.
Prompts per vertical are written to span seven categories: top-of-funnel (broad category awareness), alternatives, comparison, persona-specific, use-case, free-tier, and emerging (AI-native phrasing). The full prompt set is public — see the footer link.
Scoring
Mention rate and citation rate are reported as two separate ranks. We deliberately do not collapse them into a single “Visibility Score” in v1 — a composite would mix distinct signals and we want each to be auditable on its own terms.
What we do not claim
The MentionAudit Index does not rank the underlying AI engines, judge model quality, or claim that more visible brands are better products. The Index measures how often the AI surfaces a brand — nothing else. Buyers and brands should treat the numbers as one signal among many.
AI-content disclosure
The rankings on this site are derived from outputs produced by third-party AI models. We sample those outputs via paid APIs during the published sampling window. We do not edit model outputs before counting; we do filter common-English-word collisions on a fixed list maintained alongside this methodology.
Opt-out & takedown
Brands may request removal from public leaderboards or a methodology correction. Requests are reviewed within ten business days. Removals are honored unconditionally; corrections are evaluated against the published methodology and answered with a documented decision.
Submit a request via the contact form (topic: index-opt-out).
Versioning
Every Index ranking is tagged with the methodology version active at sample time (currently v1.0.0). A methodology change — adding or removing an engine, editing the prompt list, or changing scoring — bumps the version and starts a new comparable series. Historical rankings stay legible under the version that produced them.