GlossaryAI search
What is AI share of voice?
Also called: share of voice in AI answers, AI visibility, LLM visibility
Definition
AI share of voice is how often your brand is mentioned or cited in AI assistants' answers to a fixed set of relevant prompts, compared with competitors. It's the AI-search version of share of search.
AI share of voice, explained
There's no Search Console for ChatGPT. Assistants don't report how often you appear, and answers change between runs, users, models and days. So visibility has to be measured from the outside, by asking the same questions repeatedly and counting what comes back.
A typical method: write a prompt set of 20 to 100 questions your buyers actually ask, like "best launch platforms for developer tools" or "how to get a startup listed in directories". Run each prompt on the assistants that matter (ChatGPT, Perplexity, Gemini, Google's AI Overviews), several times each. For every answer, record whether you're mentioned, whether you're linked as a source, where you appear, and which competitors appear. Your share of voice is your mentions or citations divided by the total across you and your competitors.
Separate mentions from citations. A mention means the model named you. A citation means it linked to your page as a source, which usually comes from live search. They respond to different work: citations to crawlable, clear, well-ranking pages; mentions to how consistently you're described across the web.
Treat the numbers as a trend, not a score. Small prompt sets swing a lot, and assistants personalize and update constantly. The useful signals are direction over weeks, which prompts you never appear in, and which competitor pages keep getting cited instead, because those pages show you what the assistant is looking for.
Be wary of tools promising exact AI rankings or traffic numbers. The honest output is a sampled measurement with a stated method.
Why it matters for founders
If buyers ask assistants before they search, you need to know whether you're in those answers. Tracking a fixed prompt set is the only reliable way to see that and tell whether your work is changing it.
Example
You track 40 prompts weekly. In month one you're cited in 3 answers and a competitor in 22. Their cited pages are detailed comparison pages. You publish your own, and your citations start to climb over the following weeks.
Common mistakes
- Checking a prompt once by hand and drawing conclusions.
- Counting mentions and citations as the same thing.
- Tracking prompts nobody asks, like your own brand name only.
- Trusting tools that report exact AI traffic without explaining their method.
Related terms
- Generative engine optimization (GEO)Generative engine optimization (GEO) is the practice of making your content more likely to be found, used and cited in answers from AI systems like ChatGPT, Perplexity, Claude, Gemini and Google's AI Overviews.
- LLM citationAn LLM citation is a link or named source that an AI assistant shows alongside its answer to say where information came from. Being cited sends referral traffic and signals that your page was trusted.
- AI OverviewsAI Overviews are AI-generated summaries Google shows at the top of some search results, with links to supporting pages. AI Mode is Google's separate, conversational search experience built on the same idea.
- ImpressionsIn Google Search Console, an impression is counted each time a link to your site appears in a search result someone is shown. It measures visibility, not visits.