In a big city, AI spreads the names thin
I assumed a business in Amsterdam and one in Haarlem would play the same game with AI. Same trade, same question, same answer box.
They don't.
In October I asked ChatGPT, Gemini, Claude and Grok the same three questions for ten trades in Amsterdam, Utrecht and Haarlem. Dentists, roofers, mortgage advisers, kitchens, and so on. Questions in Dutch, web search on. Per city that's ten pages, with 112 to 114 answers in total.
Then I counted who got named in each city.
The numbers
Amsterdam: 378 businesses named. 237 of them came up in just one answer. Only 13 came up in six answers or more.
Haarlem: 251 businesses named. 126 came up just once. 23 came up six times or more.
Utrecht sat in between: 280 named, 145 just once, 22 at six or more.
Roughly the same number of answers in each city. In Amsterdam the AIs spread them over far more names. In Haarlem they kept coming back to the same ones.
Two pages side by side
Dentists in Amsterdam: 48 practices named, 40 of them only once. The most-named practice, TandartsJordaan, showed up in 6 answers.
Kitchens in Haarlem: 16 shops named. Keur Keukens showed up in 11 of the 12 answers.
Mortgage advisers in Haarlem: 14 named, and only one of them came up just once. De Hypotheekshop was named in 9 of 12 answers.
What I take from it
My read, not a measured fact: in a big city the AI has more businesses to pick from, so each answer pulls a different handful. Being named once in Amsterdam is easy. Being named again and again is hard.
In a smaller city the list is shorter. If the AI knows you, it tends to keep naming you. If it doesn't, there are fewer slots to squeeze into.
So the question isn't only "does AI know me?" It's "does it pick me over the others, every time someone asks?" Where you are changes how hard that is.
Check yours
The free scan at foreground.info asks the AIs what your customers ask, in your city, and shows who comes up.
Same question. Different city. Different game.