We asked AI to recommend local businesses. It named 9 of 60 Google Maps leaders

In November 2026 we took 60 businesses from the top of Google Maps in five cities (Toronto, London, Sydney, Chicago, Dublin) and six niches: dentists, beauty salons, auto repair, lawyers, restaurants and gyms. Then we asked an AI assistant with web search the question a real customer asks: "What is the best dentist in Toronto? Recommend a few."

Across 30 questions it recommended 156 businesses. Only 9 of our 60 map leaders (15%) were among them.

How we ran it

• In each city and niche: the business in 1st place on Google Maps (once the 2nd) and the one in 6th place.
• 30 plain-English questions to Claude with web search on, one per niche and city, on 3 November 2026.
• Answers matched to our list by website address and name.

One answer per question, one assistant, one day, small groups. Read it as a snapshot, not a ranking.

First place on the map barely helped

Google's organic results mattered more

That fits how assistants with web search work: they read search results and the pages behind them. All 60 businesses ranked first for their own name, and it did not help: customers ask about a niche, not a name.

The niche changes everything

Gyms: 6 of 10 named. Dentists, salons, auto repair: 1 of 10 each. Lawyers and restaurants: 0 of 10. In those two niches the AI answered with the biggest, most famous names in town.

A tidy website alone did not decide it

On 17 basic site checks, the businesses the AI named scored lower on average: 54 vs 67 out of 100. A good site converts visitors, but on its own it did not get a business recommended.

Two assistants, two lists

For 23 of the businesses we also had a Gemini answer from the day before. Gemini named 6, Claude named 3, and none was named by both.

Data

Aggregate table (CSV, CC BY 4.0): https://veilvex.com/data/ai-recommendations-local-business-2026.csv

Full write-up with all tables: Who does AI recommend? We checked 60 local businesses