Your position on Google no longer tells you whether buyers find your brand when they ask an AI. Measuring AI share of voice is the equivalent, in this new environment, of checking where you rank in a traditional search engine: the percentage of generative AI responses, within a defined set of questions, in which your brand appears mentioned or cited. It is also known as citation share or share of model.
Why looking at your Google ranking is no longer enough
Research from BrightEdge AI Catalyst found that brand mentions did not match in 61.9% of cases between Google AI Overviews, AI Mode, and ChatGPT for the same queries. In other words, your citation share is specific to each engine and cannot be inferred from your SEO positions. Dominating traditional Google results guarantees nothing about how ChatGPT or Perplexity presents you when someone asks them the same question.
The manual method, without spending a dime
You don’t need a budget to start measuring. The basic process is:
- Create a list of 15 to 20 real questions your ideal customer would ask about your category, not about your brand.
- Run them through ChatGPT, Perplexity, and Google AI Overviews, always in an incognito window to prevent the AI from personalizing the response based on your history.
- For each question, record whether you appeared, which other brands appeared, whether the AI cited any source with a link, and whether it was yours.
- Repeat the same exercise, with the same questions, once a month to detect trends.
This manual method has one advantage that no paid tool can replace: it forces you to read the complete responses and see, with your own eyes, how AI presents you compared with your competitors.
When it makes sense to scale with a tool
When manual tracking becomes unmanageable—with many brands, many competitors, and many engines—there are specialized platforms that automate prompt submission and aggregate your citation share for you. Tools such as Otterly.ai or HubSpot’s AEO Grader offer free or accessible entry points; options such as Profound or Ahrefs Brand Radar cover more engines and provide greater depth, with plans designed for enterprise teams.
Not every mention is worth the same
When recording your results, distinguish the level of visibility: it is not the same for AI to mention you in passing within a list of ten options as it is for it to directly recommend you as the best alternative. Note this distinction in each measurement: it will tell you much more about your actual progress than simply whether you appeared or not.
Frequently asked questions about measuring AI share of voice
How often should I measure my AI share of voice?
Once a month is enough to detect trends without generating noise. Always use the same set of questions so that results are comparable from one cycle to the next.
Why should I use incognito mode when measuring?
Because without being signed in, you prevent AI from personalizing the response based on your search history, which could give you a false positive about your own visibility.
Can I measure the traffic these AI platforms send in my analytics?
Yes. Traffic referred by ChatGPT, Perplexity, or Gemini is recorded as a referral in your web analytics, allowing you to cross-reference citation share with actual visits rather than just appearances in responses.
