Skip to main content
Share of voice in AI is the percentage of AI-generated answers to a defined set of questions in which a brand is mentioned, measured against the competitors that appear in the same answers. It answers: when a buyer asks an AI assistant about this category, how often does our brand come up compared to everyone else’s?

​
How it is calculated

Take a fixed set of representative buyer questions. Run each against an answer engine. For every answer, record which brands are named.
A competitive view divides each brand’s mentions by the total mentions across all brands, so the figures sum to 100%.

​
What makes the number meaningful

  • The query set defines the result. A set weighted toward branded or niche questions inflates the score. Use unbranded, buying-intent questions a prospect would actually ask.
  • Set size matters. Answer engines are volatile, so read many answers, not one. In Ryze each tracked prompt — up to 50 — is asked on four assistants (ChatGPT, Claude, Gemini and Perplexity), so one refresh yields up to 200 answers; compare refreshes over time rather than judging a single run.
  • Query wording changes the field entirely. Questions phrased around a platform tend to be dominated by that platform’s own properties, leaving little room for third-party brands. Questions phrased around the problem are far more open.
  • Compare like with like. Only track movement against a baseline captured with the same query set and the same engine.

​
Share of voice vs citation share

They measure different things and can move independently. Share of voice is the harder number to move and the slower to respond. Mentions come from the model’s training data and move slowly; citations come from live retrieval and respond to content — see AI visibility.