Two vendors can both report a "30% visibility" number and mean different things. Here are the precise definitions and formulas for mention rate, citation rate and share of voice, plus how to read them as one system.
GEO is a young field and its measurement language has not settled. We regularly see a provider report "visibility" that actually counts recommendations, or a client read a "citation" number that only logged a brand name in passing. Because the words sound interchangeable, the confusion rarely gets caught until a board meeting, where two charts that cannot both be true land on the same slide.
The fix is not to wait for an industry standard. It is to define each number yourself, in one sentence, and hold every report to that definition. This article sets out the three core metrics we use at AoWang, the formula behind each one, and the auxiliary metrics that tell you whether a raw number is good news or bad. If you also need the step-by-step measurement loop, see our guide to measuring GEO end to end.
Mention rate (often called visibility rate) is the share of your test questions in which the AI names your brand at all. The formula is simple:
Mention rate = (questions in which your brand appears ÷ total test questions) × 100%
Worked example: with a library of 50 buyer questions, the AI mentions your brand in 15 of the answers. Your mention rate is 30%. It does not matter where you appear, whether the description is flattering, or whether the answer linked to you. Mention rate answers exactly one question: are you in the candidate set when a buyer asks about your category?
One definitional note: a stricter reading of "mention" counts occurrences, so an answer that names you twice contributes two mentions rather than one. We recommend measuring whether a question mentions you at all for the headline rate, and counting occurrences only as a secondary detail. The danger is comparing a question-based number against an occurrence-based number and drawing a false conclusion. Mention rate is the most direct GEO result because the goal of GEO is, first and foremost, to be mentioned.
Citation rate goes one level deeper. It counts the share of questions in which the AI does not merely name you, but uses your content as an explicit source of evidence, identified by a named source or a visible link. On ChatGPT and Gemini this appears as a cited reference; on Perplexity it appears in the source list under the answer.
Citation rate = (questions where the answer cites your content as a source ÷ total test questions) × 100%
A citation matters more than a bare mention for two reasons. First, it shows the model found your content extractable enough to build on, which is the direct result of structured, fact-dense pages. Second, a cited source typically reaches the buyer directly: on Perplexity especially, readers click through sources, so a citation is also a discovery pathway. In our monitoring, citation rate is usually the metric that moves second, after mention rate, and moving it requires content and corroboration work rather than awareness work.
| Signal level | What the answer does | What it implies |
|---|---|---|
| Zero mention | Your brand never appears | You are not in the AI candidate set at all |
| Mention | Brand name appears, no source attached | Recognized entity, weak evidentiary weight |
| Citation | Your content used as a named source | Content is extractable and trusted |
| Recommendation | You are the first or suggested pick | Preferred in the buyer shortlist |
SEO taught the industry to watch rank share: how often you appear on the first results page versus competitors. In AI answers the equivalent is share of voice, which we call share of model. It is your mentions as a share of all brands' mentions in the same answer set:
Share of voice = (your brand mentions ÷ total mentions of all brands in the answers) × 100%
The difference from mention rate is the denominator. Mention rate compares you to your own possible coverage; share of voice compares you to your market. A 30% mention rate sounds weak if two competitors together hold 70% of the mentions in your category, and strong if the field is fragmented. This is why we always record which competitors appear in every answer: without the competitor row, a mention rate has no competitive meaning.
Share of voice is the metric leadership most often cares about, because it is the AI-era replacement for "how much of the conversation is mine." Track it monthly against your closest two competitors on the same question set, and the trend is what reports should show rather than any single snapshot.
The three core metrics answer different questions and should be read together, never blended:
As planning ranges, not guarantees: brands typically reach a 20-30% mention rate in about three months of consistent work and 40-50% by six months, with the leading tier being the first or recommended option on core buyer questions. We recommend aiming for the top three mentioned brands on core queries within three months and the top spot within six, measured against your own baseline. For the ongoing cross-engine execution behind these numbers, see our guide to monitoring visibility across ChatGPT, Gemini and Perplexity.
Key takeaway: Mention rate asks whether you appear; citation rate asks whether the answer uses you as evidence; share of voice asks how much of the AI conversation you own versus competitors. Define each formula once, archive the raw answers, and report all three as separate lines. A single blended "visibility" number hides which layer actually moved.
Want the vocabulary applied to your category? Start from our beginner's guide to GEO, or ask the AoWang team to benchmark your mention rate, citation rate and share of voice against your competitors.
Citation rate is the share of your test questions in which the AI answer uses your content as an explicit source, identified by a named source or a link. It is stricter than a mention rate, which only requires your brand name to appear somewhere in the answer. A citation means the model treated your content as evidence, which is a stronger trust signal.
Share of voice, also called share of model, is your number of brand mentions divided by the total mentions of every brand in the same set of AI answers. Where mention rate measures your absolute coverage, share of voice measures your position relative to competitors, which is the metric that matters most for board-level competitive context.
In practice they are often used interchangeably, but a strict definition separates them. Visibility rate asks whether your brand appears in the answer at all, regardless of how many times. Mention rate can count occurrences, so one answer that names you twice could add two mentions. Agree which definition you are using before comparing numbers across vendors or reports.
As planning ranges rather than guarantees: reach a 20-30% mention rate within three months of consistent work, and 40-50% by six months. On your core buyer questions, aim to enter the top three mentioned brands within three months and to be the first or recommended option within six months. Always benchmark against your own baseline and your closest two competitors.
Talk to the AoWang GEO team about visibility diagnosis, content optimization and AI-search monitoring for your brand.
Contact Us →