Our methodology
How we measure
AI visibility.
Six metrics that reveal exactly how AI search engines see your brand. No vanity numbers. No black boxes. Evidence you can act on.
Mention Rate
How often is your brand named at all when AI responds to buyer prompts in your category? This is the most basic measure of visibility — if you are not mentioned, you do not exist in that conversation.
Measured across 50+ prompts per category, per AI platform.
Recommendation Rate
Being mentioned is not enough. How often does AI actively recommend your brand as a solution? We distinguish between passing mentions and genuine endorsements.
Scored on a 3-point scale: not mentioned, mentioned, recommended.
Citation Rate
When AI recommends your competitors, which sources does it cite? This reveals the third-party content that shapes AI opinion in your category.
Maps the citation graph between AI responses and source content.
Competitive Share
What percentage of recommendation slots in your category does your brand capture versus each competitor? This is your share of AI voice.
Calculated as your recommendations divided by total category recommendations.
Narrative Accuracy
Does AI describe your brand correctly? Inaccurate positioning, outdated features, or confused messaging can be more damaging than invisibility.
Evaluated against your current positioning and verified factual claims.
Source Coverage
Of the sources AI trusts in your category, how many mention your brand? Source coverage is the leading indicator — improve coverage and recommendation rates follow.
Tracked across review sites, publications, comparison pages, and community forums.
Platform coverage
Version one covers four engines. An analyst runs the prompts and reads every answer. We do not add an engine to this list until it is part of the standard engagement.
- ChatGPTBuyer prompts run against the current ChatGPT product
- GeminiGoogle Gemini, including what it cites
- PerplexityPerplexity answers with their cited sources
- Google AI OverviewsThe overview shown on Google for the same questions