What is AI share of voice?
AI share of voice, also called share of model, is the percentage of AI answers in your category that mention or cite your brand versus your competitors. If 100 answers in your space name a brand and 12 of them name you, your AI share of voice is 12 percent. It is the AI-era analog of the old share-of-voice metric, moved from ad impressions to machine answers. As one way to put it, “it measures how often the machines mention you, not how often you rank.” the GEO study (Princeton, KDD 2024) showed that what a model cites can be shifted by how content is written.
The short definition
Share of voice is an old marketing idea: out of all the attention in your category, how much of it is yours. For decades it was measured in advertising, your spend or your impressions against the total in the market. AI share of voice moves that same question into the answer engines. When people ask ChatGPT, Perplexity, Google's AI Overviews, or Gemini about your category, those tools name brands. AI share of voice is simply the fraction of those mentions and citations that point to you rather than to a competitor. It is a presence metric, not a traffic metric. It tells you how often the machines speak your name when the topic is yours.
How it works
Measuring AI share of voice is a counting exercise run on a fixed set of prompts. The shape of it looks like this:
- Define the prompt set. Write a consistent list of category questions a real buyer would ask, the kind that should surface brands like yours. Hold that list steady so results are comparable over time.
- Run it across the engines. Send the same prompts to ChatGPT, Perplexity, Google AI Overviews, and Gemini, because each one retrieves and cites differently.
- Count mentions and citations. For each answer, record which brands are named and which are cited with a link. Your share is your count divided by the total brand mentions across the set.
- Track it over time. Re-run the same prompts on a schedule. The trend matters more than any single snapshot, because the engines and the field both move.
The discipline is in keeping the prompt set and the counting rules constant. Change the questions and you are measuring a different thing.
Why it matters
AI share of voice matters because it is becoming the real visibility metric as clicks decline. When answers are delivered on the results page and inside chat tools, a rising share of your audience learns about your category without ever visiting a site. In that world, ranking and even traffic undercount your true presence, while the count of how often the machines name you tracks it directly. It is also competitive in a way a raw audit score is not. Your own page can be technically excellent and you can still be invisible in answers because three competitors are the ones getting cited. Share of voice puts you on the same scoreboard as them and shows the gap in plain numbers. That makes it a target you can move, by becoming the source the answers are built from rather than just a well-optimized page nobody quotes. The background on optimizing to be that source lives in our guide to answer engine optimization (AEO).
How to check and apply it
You raise AI share of voice the same way you earn any citation, by being the cleanest, most quotable source in your category, then tracking the result.
- Build a baseline. Run your category prompt set across the major engines and record where you appear and where competitors appear instead. That gap is your starting number.
- Be extractable. Serve direct, self-contained answers in real HTML so the engines can lift and cite your passages rather than a rival's.
- Win the prompts you lose. For each question where competitors are named and you are not, study the cited pages and write a stronger, more specific answer.
- Re-measure on a cadence. Track mentions and citations over time, not once, so you can tell a real gain from noise.
You can run the prompt counting by hand, or paste your link into our AEO checker and let Brimm grade how citable your pages are in the first place, since a page no engine can extract will never earn share. When you want the full picture and the fixes ranked, run the audit at Brimm.
See also
AI share of voice rises as the link between impressions and clicks weakens, the pattern covered in what is the great decoupling. The discipline of earning those mentions in the first place is answer engine optimization (AEO). And if your share of voice is near zero because you are absent from AI answers entirely, start with why isn't my site showing up in AI search.