AI Share of Voice: How to Measure Your Brand in AI Answers
On this page
Gartner predicted in February 2024 that traditional search volume would drop 25% by 2026 as users moved to AI chatbots, and by late 2026 the AI answer layer is where an increasing share of buying decisions start. That makes a new metric necessary: AI share of voice, the percentage of your target queries where your brand is cited, mentioned, or recommended in AI answers. If you cannot measure it, you cannot win it.
What AI share of voice measures
For each query in your set, an AI answer can do one of four things with your brand:
- Cite it as a source, the strongest outcome.
- Mention it by name without a link.
- Recommend your category without naming anyone, an opportunity.
- Ignore it entirely while citing a competitor, a loss.
Your AI share of voice is the share of queries where outcomes 1 and 2 go to you rather than a competitor, plus a sentiment check: is the model describing you accurately?
Building the query set
Start with 25 to 50 queries in three buckets:
- Customer questions: the exact phrases your customers type and ask, including best [category] in [town].
- Category questions: how much does X cost, what is the difference between X and Y.
- Comparison questions: you vs your three named competitors, which [category] is best in Ireland.
The set must reflect real buying journeys, not vanity terms. Twenty queries you genuinely compete on beat two hundred you do not.
Scoring across engines
Run each query on ChatGPT, Perplexity, Gemini, and Google AI Overviews, and record the outcome per engine. Note the date, the prompt used, and any citations. Two rules keep the data honest:
- Use the same prompts every time, so readings are comparable month to month.
- Log the raw answer, because models update and you will want the evidence when a competitor disputes a trend.
Benchmarking competitors
Share of voice is relative. Score your three closest competitors on the same query set and the same engines. The output is a table: for every query, who got cited, who got ignored, and where the model described someone incorrectly. Competitor citation gaps are your work list.
Cadence and reporting
Measure the full set monthly and the ten highest-value queries weekly. Plot the trend: are you gaining or losing citations? Tie each movement to something you shipped, so measurement feeds the playbook in our guide to getting cited.
Doing it manually vs automating it
For a small set, manual measurement takes a few hours a month and a spreadsheet. At scale, it becomes a monitoring service: scheduled prompts, stored answers, scored citations, and a dashboard. We built ours exactly this way, covered in how to measure AI share of voice. One client went from invisible to consistently cited in its category within two quarters, documented in the AI search visibility case study.
Start here
Pick your ten highest-value queries, run them across the four engines today, and record the baseline. That single spreadsheet is the beginning of a GEO program. If you want the full system, our AI search visibility service includes the baseline, monitoring, and the fix-it list.
Frequently asked questions
What is AI share of voice?
The percentage of your target queries where your brand is cited, mentioned, or recommended in AI answers across ChatGPT, Perplexity, Gemini, and Google AI Overviews, relative to your competitors.
How is AI share of voice different from rank tracking?
Rank tracking follows your position in a list of pages. AI share of voice follows whether a model names your brand when it answers. A page can rank first on Google and never be cited.
Which AI engines should I track?
At minimum ChatGPT, Perplexity, Gemini, and Google AI Overviews. Add Copilot and local assistants if your audience uses them. The set should match where your customers actually ask questions.
How many queries should I track?
Start with 25 to 50: your top customer questions, category queries, and competitor comparison queries. Precision on a small set beats noise on a large one.
How often should I measure?
Monthly for the full set, weekly for the ten queries that drive revenue. AI answers change as models update, so a trend line matters more than any single reading.
Can I measure AI share of voice myself?
Yes for small sets: ask the same prompt on each engine, record who is cited, repeat monthly. At scale, monitoring services and APIs remove the manual work.