AI Visibility vs. AI Citation: How the Two Relate
Explains, with one computed statement instead of two disconnected percentages, whether AI engines know your brand, whether they cite your actual pages as a source, and why those can genuinely disagree.
How it works
These measure two different things. AI Visibility (Mention Rate) sends generic, brand-awareness-style prompts to Google AI Overviews, ChatGPT, and Gemini, prompts a customer might ask before they know which brand to pick, and checks how often your brand name gets mentioned anywhere in the answer. AI Citation Score (Citation Rate) is narrower and more specific: for a set of your real target keywords, it checks whether Google actually shows an AI Overview at all, and if it does, whether one of the cited source links points to a page on your site.
Being mentioned by name and being cited as a source are genuinely different outcomes. An AI engine can know your brand exists (from training data, reviews, general web presence) without ever treating one of your specific pages as a citable source for a specific query, that's usually a content-format problem: the page exists but isn't structured in a way the AI extracts and cites. The reverse also happens: a page can already be cited as a source while brand-awareness prompts still rarely name the brand itself, that's a brand-recognition gap, not a content gap.
The relationship reading you see (the "How your two AI scores fit together" card) is not written freehand by an AI given the two numbers. It's a fixed decision tree with four possible outcomes, both low, both healthy, mentioned-but-not-cited, or cited-but-not-mentioned, and the specific numbers are filled into whichever template matches. The reasoning is deterministic, only the sentence-level phrasing pulls from a template written in advance.
What this means for you
Two different reasons can produce the same-looking "0% citation" number, no AI Overview ever appeared for your keywords yet (nothing to be cited in), or AI Overviews appeared and cited someone else. The card tells you which one you're looking at instead of leaving you to guess.
Known limitation
Both checks run a fixed, small sample of prompts/keywords per check (not your full keyword universe), and AI answer engines are non-deterministic between runs, the same prompt can occasionally return a different answer minutes later. Treat a single check as a real sample of current behavior, not a guaranteed, permanent state.