More and more customers begin their decision by asking an AI assistant instead of searching Google. The natural question is: does my brand show up in those answers? Measuring it is not as simple as checking your position in a search engine, but it is both possible and necessary. Here is how to know whether AI mentions you and what you should be measuring.
Why there is no ranking like Google
In traditional search, your site occupies a spot in a list of links. In generative engines there is no list: there is a synthesized answer in which your brand appears, is cited, is described… or simply does not exist. That is why AI visibility is not measured with a ranking, but with how often and how the models name you when someone asks about your category.
The manual method: ask and observe
The first step, which you can do right now, is to type your real commercial questions into ChatGPT, Gemini, Perplexity and Claude: "which company do you recommend for X?", "best providers of Y", "how do I choose Z". Note whether your brand appears, in what order versus competitors, and whether what it says about you is correct. It is revealing, but it has limits.
Why the manual method is not enough
AI answers vary: they change between models, between regions, between versions and sometimes between one query and the next. Asking once does not tell you whether you appear 5% or 80% of the time. To make decisions you need to measure systematically: many prompts, repeated over time, across each engine and in your country.
What you should measure
- Share of Model Voice (SoMV): the percentage of relevant prompts in which you are named. It is the generative equivalent of your market share inside the answers.
- Citation Rank: your position versus competitors when several brands appear in the same answer.
- Factual accuracy: whether the model describes your prices, services and coverage correctly.
- Sentiment: the tone with which it presents you.
- Hallucination index: how often it invents data about you.
How to monitor it seriously
Professional monitoring defines a set of commercial prompts per market, runs them periodically across the leading AI engines and records how each metric evolves. That way you can see, week by week, whether your presence is rising, where competitors beat you and which prompts still do not mention you. That measurement is the foundation of the optimization work: what is not measured cannot be improved.
From measurement to action
Measuring is the starting point of the GEO-5 method: first a diagnosis of your real presence, then building your entity record and citable content, and finally a continuous sweep to reach the top spot on the prompts that matter. Without measurement you work blind; with it, every improvement is verifiable.
How a good visibility dashboard looks
A useful dashboard is not a pile of screenshots, but a view that evolves. It should show your Share of Model Voice by engine and by prompt family, its trend week by week, and a clear comparison against your main competitors. When the number goes up, you want to know why; when it drops, you want an alert.
Beyond the appearance rate, a good dashboard records the order in which you are cited, the tone used to describe you, and the cases where the model invents data about you. With that reading you can prioritize: attack first the high-intent commercial prompts where you do not yet appear, fix the wrong descriptions that damage your reputation, and reinforce the sources the model is using to talk about your category. In practice, that dashboard turns a vague feeling — "I don't think I show up in AI" — into a concrete, measurable plan of work.
Start today
Run a manual test with five real questions from your industry across three different engines. If your brand does not appear — or appears poorly described — you already have the signal that you need an AI visibility strategy. The next step is to turn that intuition into a metrics dashboard you can track over time.
Frequently asked questions
How do I know if my brand appears in ChatGPT or Perplexity?
The quick way is to ask them directly the commercial questions in your industry and observe whether they name you, in what order and whether the information is correct. Because answers vary across models, regions and versions, to make decisions you need to measure systematically: many prompts, repeated over time and across each engine, tracking metrics like Share of Model Voice and Citation Rank.
How often should I measure my AI visibility?
Ideally continuously (weekly or monthly), because models and their answers change frequently. A single measurement is a snapshot; the value is in the trend.
Is it useful to appear if the AI describes my company poorly?
Not entirely: appearing with wrong data can be worse than not appearing. That is why, beyond presence, you also measure factual accuracy and how often the model hallucinates about your brand.
Informational article by Petri Heil · GEO & SEO. Want to know how your brand shows up in AI today? Request a diagnosis.