How AI assistants choose which local businesses to recommend

Why this matters now

For twenty years, being found locally meant being found on Google. That’s changing. In BrightLocal’s 2026 consumer survey, 58% of people said they had already used AI to find or get a recommendation for a local business.

The difference is what the customer sees. Google shows a list and a map, and people browse. An AI assistant gives a short answer with a handful of names, and people usually pick one of those. If you’re not in the answer, there’s no second page to scroll to.

In the UK, ChatGPT is by far the most used assistant, with about 8 in 10 chatbot visits according to StatCounter. Gemini is a distant second. That’s why it matters most what ChatGPT says about your area.

How an AI answer is made

When someone asks “which nail salon in Clapham would you recommend?”, a modern assistant roughly does this:

1. Searches the web, much like you would, using a search engine behind the scenes.
2. Reads the top results: business listings, review sites, local articles, sometimes business websites.
3. Looks for agreement. Businesses that appear in several trusted places, with good reviews, stand out.
4. Writes a short answer, usually naming three to five businesses, often with a reason for each.

Where AI looks

Assistants often show the sources they used. Across the questions we ask, the same kinds of sites come up again and again:

• Google Business Profiles and Google reviews. Gemini leans on Google directly.
• Bing and Apple Maps listings. ChatGPT has historically used Bing for local search, which is why Bing Places matters more than most owners think.
• Review and booking sites for your trade: Tripadvisor, Treatwell, Checkatrade, Yelp and similar.
• Local articles and “best of” lists from newspapers, blogs and city guides.
• Your own website, if it clearly says what you do and where.

Why AI picks your competitors

When we compare businesses AI names with businesses it skips, the differences are rarely about quality. They’re about information:

• More reviews. One study of 10,000 local businesses found AI-recommended businesses averaged 133 Google reviews, against 11 for those AI ignored.
• Complete listings. Clear categories, services, hours and descriptions make it easy for AI to match a business to a question.
• Consistency. The same name, address and phone number across many sites makes a business look established.
• Mentions. A business named in a local “best of” article gets repeated in AI answers.
• Clear words. If customers ask for “gel nails” and your listings only say “beauty”, AI may not connect the two.

The same study found 41% of local businesses are never recommended by ChatGPT, even when it knows they exist.

Why answers change

Ask the same question twice and you may get different names. Ask three assistants and you’ll almost certainly get three different lists.

This happens because each assistant searches again every time, reads slightly different pages, and writes its answer fresh. Answers can also be shaped by:

• Wording. “Best”, “cheap”, “near the station” and “open late” all change the answer.
• Location. Assistants may use your rough location to decide what “near me” means.
• Your history. If you’re signed in, some assistants remember your past chats and preferences.

So the useful question isn’t “did it name me once?”. It’s “how often does it name me, across many questions, compared with my competitors?”.

How to test it yourself

You can get a rough idea in five minutes:

1. Open ChatGPT in a private browser window, signed out if you can.
2. Ask the way a customer would, without your business name: “Which [your type of business] in [your area] would you recommend?”
3. Try two or three versions: “best”, a specific service, “near [landmark]”.
4. Note who’s named, and click the sources to see where the answer came from.
5. Repeat in Gemini, and again next month.

Or let us do it: the Echolisted free check runs 40 customer searches across three assistants in about two minutes, scores the result and shows every answer, with a button to open each question in ChatGPT yourself.

What you can do

The good news: the same work that helps on Google Maps helps with AI. In order of impact for most local businesses:

1. Complete your Google Business Profile: right category, services with prices, hours, description, photos. How to do it.
2. Claim Bing Places and Apple Business Connect. They’re free and often ignored, and Bing feeds ChatGPT.
3. Get more reviews, steadily, by asking every customer. Reply to them all.
4. Fix inconsistent listings on the review and booking sites that appear in your sources.
5. Describe what you do in customers’ words, on your profile and website: the services people actually ask for, and the area you’re in.
6. Earn a local mention: an article, a local guide, a business association listing.

Myths to ignore

• “You can pay ChatGPT to recommend you.” There’s no way to buy a place in an AI’s recommendations. Be wary of anyone who says otherwise.
• “You need a special AI file on your website.” Clear information in the places AI already reads matters far more than any single file.
• “AI only recommends big chains.” For local questions, assistants often name independents, as long as they’re well described and well reviewed.
• “If I’m top on Google, I’m fine on AI.” Often, but not always. AI reads more sources and names fewer businesses, so check.

Measuring it properly

Because answers vary, a good measurement needs:

• Many questions, worded the way different customers would ask
• More than one assistant, weighted by how many people use each
• No business name in the question, so the answer isn’t nudged
• A comparison with your competitors
• The same test repeated over time

That’s exactly how the Echolisted score works. See what the check includes.

Sources

• BrightLocal: Consumer Survey 2026, as published on its Local AI Visibility page
• StatCounter: AI chatbot market share, United Kingdom (September 2026)
• Insites: What it takes to rank in AI (April 2026)
• Whitespark: review sources for ChatGPT local results