What is AI Share of Voice?
Your brand's percentage of mentions across AI answers for a prompt set, versus competitors — the market-share metric of AI search.
Definition
AI share of voice is the competitive metric of AI visibility: across a defined set of buyer prompts and a period of sampling, what percentage of brand mentions belong to you versus each competitor? If assistants answering "best X tools" name competitors 60 times and you 15 times across runs, your share of voice for that intent is 20%.
Why it matters
Averages hide the race. Your mention rate can rise while a competitor's rises faster — share of voice catches it. With AI answers naming only 3–4 brands and the top domains hoarding two-thirds of citations, visibility is zero-sum per prompt: every slot you gain, someone loses. Share of voice per intent (discovery, comparison, alternatives) shows exactly where you're winning shelf space and where a competitor owns the conversation — which is what makes it the board-level metric for GEO programs.
Frequently asked
How is AI share of voice calculated?
Run a fixed prompt set across assistants repeatedly; count brand mentions (optionally weighted by position and sentiment); divide yours by total. Consistent methodology run-over-run matters more than any absolute number.
What's a good share of voice?
Depends on category concentration. More useful than a benchmark: your trendline versus the top competitor's, per campaign intent. Rising share in comparison prompts predicts pipeline better than any single mention.
Related terms
- Schema MarkupStructured data vocabulary (schema.org) describing a page's meaning to machines — entity hygiene for search engines and AI pipelines.
- Semantic SearchSearch by meaning rather than keyword matching — powered by embeddings, it's how retrieval systems decide what's relevant to a prompt.
- AI CitationsThe linked sources an AI assistant credits when generating an answer. Being cited is the AI-era equivalent of ranking #1.
- AI CrawlerBots operated by AI companies — GPTBot, ClaudeBot, PerplexityBot, Google-Extended — that fetch web content for model training or live answer retrieval.