Ai Search

AI Share of Voice: The Metric Behind AI Citations

AI share of voice is the percentage of AI-generated answers, across a fixed set of buyer prompts, in which your business gets named or cited. It is the answer-engine version of an old search metric: instead of your slice of the clicks for a keyword, it measures your slice of the recommendations an AI model hands to a buyer who asked for one.

The number only means something against a defined prompt panel and a defined field of competitors. Take a panel of twenty or so buyer questions, run each one through ChatGPT, Perplexity, Gemini, and Google AI Overviews, then count how often every business surfaces. Your share of voice is your appearances divided by the total across that panel. A single answer rarely names more than a few businesses, so this metric behaves far closer to zero sum than ten blue links ever did.

What the metric actually counts

Share of voice is a ratio, and a ratio is only as honest as its denominator. Three inputs define it. The prompt panel is the specific set of questions a buyer would actually type, phrased the way a buyer phrases them, not the keywords you wish you ranked for. The platform set matters because ChatGPT, Perplexity, Gemini, and Google AI Overviews each retrieve and cite differently, so a single blended score hides more than it reveals. And the surface you count matters because a linked citation and a plain-text mention are not the same event.

Prompt panel: the fixed set of buyer questions you score against, held steady so the figure stays comparable over time.

Platform set: each answer engine scored on its own, because a source Perplexity leans on may be invisible inside ChatGPT.

Mention or citation: whether you count being named in the answer text, being linked as a source, or both, decided before you measure, not after.

Time window: answers drift, so any share-of-voice figure is a snapshot of a date range, never a permanent score.

Mention versus citation

This distinction decides what you are really measuring. A citation is a source the model retrieved and linked, the footnote a buyer can click to verify. A mention is the model writing your business name into the answer itself, with or without a link. The two come apart constantly. A model can recommend a business it never links, and it can link a page without ever saying the brand out loud.

For most service businesses, mentions drive the decision, because the buyer reads the name and acts on it. Citations drive durability, because a linked source is one the model can keep returning to and one you can independently check. A serious share-of-voice measurement tracks both surfaces and never pretends they are interchangeable.

Why it is not a keyword ranking

Traditional share of voice summed your visibility across a keyword set, where positions one through ten each carried a known, roughly fixed value. AI answers have no positions one through ten. The model reads several sources, synthesizes one response, and names a short list. You are either on that list or you are not, and there is no second page to drift onto.

The contrarian consequence is that breadth beats a single trophy. Ranking first for one phrase matters far less than being the source a model reaches for across many related prompts. A business cited lightly but almost everywhere will out-earn a business that owns one keyword and is absent from the rest of the conversation. Share of voice is the metric that makes that breadth, or the lack of it, visible.

How you grow your AI share of voice

You raise AI share of voice by becoming the source a model can both find and trust across your whole prompt panel. That is the same work that earns individual citations, applied at the scale of a topic instead of a single page. There is no payment that places you inside an answer, and anyone promising a guaranteed spot is selling something that does not exist.

Entity clarity: make it unambiguous what your business is, where it works, and what it does, so a model can resolve you as one consistent entity.

Content structuring: answer the real buyer questions in clean, extractable prose a model can lift without guessing at your meaning.

Crawler access: confirm that the AI crawlers you want are allowed to reach your pages, because a page a model cannot fetch cannot be cited.

Coverage: publish across the full set of prompts in your panel, not only your one headline topic.

Re-measurement: score the panel again on a cadence, because a number with no re-check is a guess dressed up as data.

Ninetyfifth Marketing treats this as ongoing work on a published monthly program, because answer engines keep changing and a one-time push decays. The method is the product: the panel, the structuring, the crawler checks, and the repeated measurement. We sell that process and those deliverables, never a promised rank and never a promised result.

Measuring it honestly

AI answers are non-deterministic. Ask the same question twice and the wording, and sometimes the names, will shift. A credible share-of-voice number accounts for that by sampling each prompt more than once, holding the panel and the platform set constant, and reporting a range instead of a single hero figure. Treat any tool that hands you one clean number with no stated method as a marketing claim, not a measurement you can act on.

Questions

How is AI share of voice different from keyword share of voice?

Keyword share of voice measured your slice of ranked positions across a keyword set, where positions one through ten each had a known value. AI share of voice measures how often your business is named or cited inside AI-generated answers, where there are no ranked positions at all. An answer names a short list or it does not, so breadth across many prompts matters more than owning one phrase.

What is a good AI share of voice score?

There is no universal benchmark, because the number depends entirely on your prompt panel, your competitor field, and which platforms you count. What matters is the direction over time against a panel you hold steady. A score only becomes useful once you can compare it honestly to an earlier measurement of the same prompts on the same engines.

Can you pay to increase your AI share of voice?

No. There is no ad slot or fee that places your business inside an organic AI answer. You grow share of voice by being findable and citable across your prompt panel, through entity clarity, content structuring, and crawler access. Anyone offering a guaranteed spot inside ChatGPT or Perplexity answers is describing something that does not exist.

How often should AI share of voice be measured?

On a regular cadence, because answer engines are non-deterministic and change their sources often. A single reading is a snapshot, not a trend. Holding the same panel and platforms and re-scoring on a schedule is the only way to tell whether your visibility is actually moving.