Contents5
In short
AI answers vary with the language of the question and the location of the user.
Being cited in French in Paris says nothing about your visibility in English in Dubai.
Tracking is done market by market, across several assistants, and read as trends.
GEO has a geography too
GEO, generative engine optimisation, aims to get you cited in the answers of assistants: ChatGPT, Gemini, Claude, Perplexity, Copilot or Google’s AI Overviews. It is often spoken of as a single ranking. It is not one.
As with Google results, which change from one city to the next, an assistant’s answer depends on the context of the question. When it searches the web to answer, it queries sources that are themselves localised. And when the user asks for “an agency near me” or “a supplier in my country”, the location goes straight into the answer.
Three reasons the answer varies
What makes an AI answer vary
- Language
A question in English draws on English-language sources.
- Location
The user’s position steers local recommendations.
- The sources of the moment
The underlying web search changes with the market.
Language is the most underestimated factor. A company with a strong presence in French-language press and directories can be cited every time in French and ignored in English, even by the same assistant, for an identical question. If your customers search in English, that is the check that counts.
On top of this comes a variability specific to AI: asked twice, the same question does not always get the same answer. Hence the rule we apply in our protocol for measuring AI visibility: we repeat the questions, and read citation frequencies rather than a single result.
Key takeaway
No single assistant is the whole market
Your customers do not all use the same one. Tracking a single assistant, whichever it is, is like measuring a single district of a city. Tracking covers several assistants, and the share of each is checked in your own visits.
How to track AI visibility by market
The principle is the same as tracking rankings city by city, adapted to AI:
- Set the questions.
The ones your buyers ask, worded in each market’s language.
- Locate the checks.
Ask these questions from the countries and cities where your customers are.
- Repeat.
Several times per cycle and across several assistants, to smooth out the variability.
- Read.
Who is cited, how often, with which sources, and how it changes.
The volume climbs fast: questions, languages, locations, assistants and repetitions multiply each other. It is mass processing, automated, whose results AI sorts (a competitor appearing in a market, a source that comes back in every answer) before an expert decides. We run it alongside the Google checks in our rank tracking on Google and in AI.
From the field
A client was regularly cited by the assistants when they were asked in French. Its buyers, based in the Middle East, asked their questions in English: in those answers it never appeared, and competitors present in English-language sources did instead. The work focused on those sources, not on its French site.
What drives progress in a given market
The levers are those of GEO, applied market by market: pages in the market’s language that clearly answer the questions asked, a presence in the sources the assistants consult for that country, and consistent information from one language to another. We detail these levers in how to get cited by ChatGPT and the other assistants.
When a market needs pages in numbers (by city, by service, by language), programmatic SEO produces them from data specific to each one, so that they serve both Google and the assistants.
Frequently asked questions
Can you know exactly where an assistant gets its answer from?
Sometimes, when it shows its sources. Otherwise, you infer it by noting which sources keep coming back in the answers for a given market.
Do you need different content for AI and for Google?
No. A clear page, in the right language, that answers the question of its market serves both.
Jim · founder of Brain Plus AI
Also in seo and ai visibility
Practical guide8 min
Measuring your visibility in AI: an honest protocol
No tool gives you a Search Console for assistants. Three readings repeated every month show whether a site is read, cited and visited by ChatGPT, Claude, Perplexity or AI Overviews, and what each measure does not tell you.
Understand4 min
Why Google shows different results in Paris and Dubai
Location, language, device, the context of the query: the same search does not return the same page depending on where it is typed. What this changes for an SEO strategy, and why a "national" ranking tells you almost nothing.
Understand3 min
Tracking Google rankings city by city: method and pitfalls
A national average ranking reassures and misleads. What you measure when you track rankings by city, language and device, how to read the gaps, and the pitfalls that cost a lot for little information.