Do AI Assistants Recommend Local Businesses? How ChatGPT and Gemini Pick a Nearby Business

Yes, and they name far fewer businesses than a search results page shows. Ask an assistant for a plumber, a dentist or somewhere to eat nearby and you will get two or three names, not ten links. The shortlist is short, which makes the difference between being on it and being near it larger than any equivalent gap in traditional search.

What happens when someone asks

No engine publishes how it picks. What is observable, by running the questions repeatedly and reading what gets cited, is that local recommendations track established business information rather than website quality.

The evidence that appears to carry weight is the evidence that is structured, verified and repeated: business profiles with complete categories and hours, reviews with real text, and name, address and phone details that agree across every source. Your website participates, but as confirmation rather than as the primary case.

This surprises people who have invested in their site. It is consistent, though, with what the engines are trying to do. Recommending a business to a stranger is a risk. Models manage that risk by naming businesses they can describe confidently, and confidence comes from repetition across independent sources, not from one website saying good things about itself.

Why the shortlist is so short

A map shows ten results and lets the customer scroll. An assistant names three and stops. If you are the fourth-best-known business in your area, the map still gives you a position and the assistant gives you nothing.

The practical consequence is that narrowing often beats competing. Being the clear answer to "emergency dentist open on Sunday in this town" is achievable. Being one of the three best-known dentists in a city is not, for most practices. Specific categories, specific services and specific review language are how a smaller business gets named at all, and that specificity is a positioning decision before it is a marketing one.

The four things that get a local business named

1. A complete, specific business profile

Precise primary category, real secondary categories, every service listed individually, a description stating what you do and who you serve, accurate hours including holidays, genuine photographs. This is the most trusted structured description of your business that exists, and it is free.

2. Reviews with content, arriving steadily

Volume helps, recency helps more than most people expect, and text matters most. A review that names the service and the outcome supplies a usable sentence. A silent five stars supplies a number. Ask customers for a sentence about what you did rather than for a rating, and reply to every review, because an answered review reads as an active business to a person and leaves more text on the page for everything else.

3. Details that match everywhere

Name, address, phone, category, identical across your site, your profile and every directory. Contradictions do not merely fail to help; they actively lower the confidence with which anything will name you. This is boring, unglamorous work and it is the single most common thing found broken during an audit.

4. Third-party corroboration

Directory listings in the platforms your category uses, local press, community and association pages, partner sites. This is what builds the association between your name, your service and your place, and it is the slowest and most durable of the four.

What your website should contribute

Not volume. Three specific things.

• A page per service, each one answering the question a customer would ask about that service, with the answer in the first paragraph.

• Unambiguous location information: the areas you serve, named, plus hours and contact details that match your profile exactly.

• An identity sentence: what you are, where you operate, who you serve, stated plainly. This is what makes attribution possible when the model has already decided to use your information.

That last one is the difference between being read and being named, and it is the same fix as the cited-but-not-mentioned problem that affects every kind of business, not just local ones.

How to test it this week

1. Write ten questions a customer would ask: your service plus your town, your service plus a neighbourhood, your service plus a qualifier like open now, affordable, or emergency.

2. Add three questions with no location, to see whether you surface for the category at all.

3. Run all of them in fresh sessions in ChatGPT, Gemini and Perplexity.

4. Record two columns: named in the answer, and cited as a source.

5. Note who is named instead of you, then open their profile and compare it with yours field by field. The gap is usually visible in under a minute.

Repeat monthly with the same questions. The method is identical to the standard AI visibility audit, narrowed to local intent.

One job, not two

The conclusion worth taking from all of this is that local visibility and AI visibility are not separate programmes. The profile that wins Maps is the profile an assistant quotes. The reviews that persuade a human are the sentences a model builds a recommendation from. The consistent business data that fixes your Maps listing is what makes a model confident enough to name you.

Markty AI covers both halves: it scans ChatGPT, Gemini and Perplexity four times a month to measure whether you are named, and it polishes your Google Business Profile, tracks Google Maps and Trustpilot reviews and drafts replies in your brand's tone so the underlying evidence keeps improving.

If you want the full picture, start with what decides local visibility and then measure the AI side. They are the same audit run twice, from two directions.