This article is part of our guide on why your website is invisible to AI.
Your network has sixty locations. In Lyon, ChatGPT names your clinic first and puts it at the top of the map. In Nantes, it names two independents and a competing chain. In Lille, your branch is on the map, eighth out of ten, and the text never mentions it.
Nobody at head office knows any of this. Your local SEO reports cover Google Maps rankings. Your brand tracking covers national questions. The answers your customers read when they ask an AI for "a good physiotherapy clinic in Nantes" fall between the two.
This article is for brands that run a network: retail chains, franchises, networks of agencies, clinics, gyms or dealerships. For them, local AI visibility is not one score. It is one result per city, and the cities do not move together.
Do people really ask AI engines for local recommendations?
They do, and the questions look like the ones your local teams hear every day. "Best physiotherapy clinic in Lyon for a sports injury." "Car rental near Bordeaux station, open on Sundays." "Which real estate agency in Toulouse knows the Carmes neighbourhood?"
The AI engines answer with names. Not a list of ten blue links to compare, but two or three businesses with a reason for each. Whoever is named is often the one called.
For a single shop, that is one answer to watch. For a network, it is one answer per city and per service, and each one has its own winners: a local independent here, a competing chain there, your own branch somewhere else.
Why is every city a separate battle?
Because the AI builds each local answer from what it finds about that city.
On a national question, your brand's reputation does most of the work. On a local one, the engine looks at the businesses around the place named in the question: their listings, their ratings, their review counts, what local media and directories say about them. Your national brand awareness helps, but it does not put your Nantes branch in the Nantes answer.
That is why networks see such uneven results. A location with a complete listing, many recent reviews and a page about its service in the city tends to show up. The branch next door, with a half-filled profile and a generic store-locator page, does not, even though both carry the same brand.
There is a second trap. A question that names no place gets a map built around wherever the person asking happens to be. To measure a network properly, the question has to name the city.
What does ChatGPT show on a local question?
There are two results hiding in a local AI answer, and most networks only look at one.
When a question carries local intent, ChatGPT writes its paragraph and attaches a map of businesses next to it, around ten of them, each with a rating and a review count. Google AI Overview does the same with a shorter local pack. Perplexity, Gemini, Google AI Mode and Claude answer in text only.
The map and the text are separate results, and you can have one without the other:
- Named in the text, absent from the map. The engine has something to say about your location but cannot find or match its listing. That is a listing problem: your Google Business Profile, its category, the consistency of name, address and phone across your site and directories.
- On the map, absent from the text. The engine knows the location exists and places it, but has nothing to say about it. That is a content problem: nothing on the web explains what this location does well enough for the AI to write a sentence about it.
Merged into one visibility score, the two cancel out and you cannot tell which one each city has. And on a local question, the map is often what the customer looks at: the ratings and the review counts in the panel, not the paragraph above it.
How do you read your AI visibility city by city?
By treating each city as a zone, with its own numbers. Mentionable's local visibility does this from the scans you already run.
On every scan of your tracked questions, Mentionable reads the business map that ChatGPT and Google AI Overview return. Each place a question names becomes a zone, and the coverage by zone shows, for each one, your brand's presence in its maps, your rank and the business that leads. Any of your locations that Mentionable recognizes as your brand, by your domain or your name, counts as you. A verdict tells you where you are already visible and where there is a zone to win.
Next to it, you see who occupies the maps over the period, with their average rank and rating, and the questions where your brand is named in the text without being on the map, or the reverse.
One limit to keep in mind: this is a network-level view. Mentionable reads where your brand stands in each zone, not a separate score for each of your locations.
Mentionable also lists the cities near your zones where the AI maps show your competitors and none of your questions asks about them. One click adapts your local questions to that city, and it joins your tracked zones at the next scan.
For a network, the setup is simple: one question per city and per service that matters to the business. Questions belong to a country and are asked from there, in its language, so a network present in several countries reads each one country by country.
What does a network fix in the cities it is losing?
The answer depends on the zone, which is why a single national action plan rarely works. Mentionable's Action plan builds the list from your data, for each zone to win in your catchment area.
The listings first. Where your brand is cited in the text but missing from the map, the engine cannot find or match your listing there. The plan points at your Google profiles: verified, with name, address and phone matching your site and the directories. When the leaders of a map are listed under a category you do not carry, it suggests adding it as a secondary category. The Google Business Profile guide covers the details.
Then the reviews. When the top 3 of a map have far more reviews or a better rating than the profile Mentionable last saw for your brand on a map, the plan gives the gap in numbers. Review volume weighs in which businesses the AI shows, so the network-wide review policy matters as much as any single location.
Then a page per city. When your brand is on the map but the AI does not name it in the text, the plan suggests a page about your service in that city, so the engine has something to cite. A store-locator entry with an address and opening hours rarely does the job. Pages to create arrive drafted in your brand voice, with the local facts only your team knows highlighted to complete, never invented.
Then the local sources. Each zone to win comes with the local sources AI engines cite there: directories, local media, rankings. You see which ones name the leader and not you. From the Influence plan, Mentionable's outreach campaigns contact the sites AI engines cite on your questions, 200 per month.
Each zone to win reads as a short plan: Google profile, service page, sources to get, with its progress. When you mark a step done, Mentionable measures your brand's presence in that zone's maps before and since, so head office can see which cities moved.
Who owns this inside a network?
This is usually the real blocker. Local listings sit with operations or the franchisees. The website sits with marketing. Reviews sit with each location. Nobody owns "being recommended by ChatGPT in Nantes".
A zone-by-zone view helps split the work without losing the picture. Head office reads coverage across all cities and decides where to push. Each action lands with the team that can do it: a listing fix for operations or the franchisees, a city page for the content team, a local source for whoever handles partnerships. The Looker Studio connector and the weekly email carry the same numbers to whoever needs them.
What does it cost to track a whole network?
Reading the map costs nothing extra. It comes from the scans you already run on your tracked questions, on every plan, and the local view appears on its own once your questions start returning maps.
What a network needs is enough questions to cover its cities and services. Presence tracks 50 questions a day, Influence 100, Reference 150, with unlimited users. For more cities, several countries beyond your plan or several brands, Mentionable sizes a custom setup with your team.
Where to start
Pick ten cities that matter to the business, a few where you know you are strong and a few where you suspect you are not. Write one question per city for your main service, naming the city. Within a few days of scans, you see which zones you hold, which ones a competitor leads, and why.
The 7-day trial needs no credit card. Enter your URL, keep the questions that matter, add your cities, and read your network the way your customers' AI reads it.
If you would rather hand the project over, Mentionable Agency goes further than the tool: its team builds the strategy, writes the content and gets the mentions with its own methods, and the tool measures your visibility in AI answers, city by city, before and after.
Related articles
- Your Competitors Are Cited by AI Instead of You: how to see it and what to do about it.
- Why Your Website Is Invisible to AI: the six root causes and how to fix each one.
- Google Business Profile and AI visibility: what your listing changes in Gemini and Google's AI answers.
