What AI search actually sees when someone asks about a local business like yours
AI assistants build their answers about your business from public sources you can check and fix. Here is where they look and how to test what they say today.
- AI assistants pull answers about local businesses from public sources: your website, review sites and online listings.
- You can test what an AI says today by asking it the same question a customer would ask.
- The fix is ordinary: correct hours and services on your site, current listings, and a steady stream of reviews.
- Recheck every month or two, because AI answers change as the sources change.
An owner in Columbus asks ChatGPT for a good HVAC company and watches it recommend a shop across town. A clinic in Toledo types a symptom into Google and the AI summary lists two practices, neither of them theirs. A restaurant owner in Cincinnati asks where to get dinner near a neighborhood and the answer names four competitors. The AI is not choosing a favorite. It is reading what the public record says, and the public record may be stale, thin or wrong.
This is the new version of an old problem. Twenty years ago a wrong phone number in the phone book cost you calls. Today a wrong or missing fact in an AI answer costs you the customer before they ever reach your site. The good news is the same as it was then: the record is public, you can read it, and you can change most of it yourself.
Where the answers come from
AI assistants do not have a private file on your business. When someone asks for a plumber, a dentist or a place to eat, the assistant assembles an answer from sources anyone can see. Google describes its AI search features as working across "the world's information," and its AI Overviews give a snapshot of key info with links to explore the web, according to Google's own Search page. In practice, that means three kinds of sources carry most of the weight.
- Your own website. The pages that state your services, hours, service area and prices. AI systems read these closely, and inconsistencies between pages cause inconsistencies in answers.
- Review sites. Google Maps reviews, Yelp and industry-specific review sites. The pattern of what customers say shows up in what an AI says about your quality, your wait times and your specialties.
- Listings and directories. Your Google Business Profile, industry directories, chamber of commerce pages and local media mentions. These confirm basics like address, hours and category.
The practical consequence is simple. If your hours are wrong on your Google Business Profile, the AI will likely repeat the wrong hours. If nobody has left a review in two years, the assistant has little to say about you and plenty to say about the competitor with forty recent reviews. The AI is a mirror, not a judge. It reflects the public record.
Test what the AI says today
The test takes fifteen minutes and requires nothing but a free account on an AI assistant and a browser. Ask the questions a customer would actually ask, in the words a customer would use. Owners tend to test with their own business name, but customers rarely search by name. They search by need and place.
- 01Ask by need, not by name
In ChatGPT, type a customer's real question, such as "who fixes water heaters in Westerville" or "family dentist near me in Parma."
- 02Check Google's AI answers
Run the same search on Google and read the AI Overview at the top of the results, if one appears.
- 03Note where you appear
Write down whether your business shows up, in what position, and which competitors appear instead.
- 04Ask follow-up questions
Ask the assistant about hours, prices, services or how to book, and note where the details come from and whether they are right.
- 05Search your business name directly
Type your exact business name and city to see what the assistant says when it knows it means you.
- 06Record the errors
List every wrong or missing fact you found: hours, services, service area, phone number, whether it knows you exist at all.
Expect different answers from different assistants and different results on different days. The sources change, the models change, and answers that name you this week may not next week. That is why the fix is a habit rather than a one-time project.
What to fix first
Start with the facts, because they are the cheapest to correct and the most damaging when wrong. Hours, address, phone number and service area should match everywhere: your website, your Google Business Profile and any directory that lists you. An AI assistant that finds three different phone numbers has no way to know which is right, and it may quietly skip you rather than guess.
Next, look at the pages on your own site. A page that clearly states what you do, where you work and who you serve gives an assistant material to quote. A page of slogans and stock photography gives it nothing. If you serve three towns in Summit County, say so in plain text. Assistants read words, not implications.
Reviews deserve real attention, not because of a score alone but because they are the only public record of what customers actually experienced. An assistant asked for the best place for a specific need leans on the pattern in those reviews. Asking satisfied customers for an honest review is ordinary business, and it feeds the same record the AI reads.
How often to recheck
Once a month is plenty for most owners, and the check itself is the fifteen-minute test above. The stakes are higher after a change: a new location, new hours, a dropped service or a renamed business should send you back to the assistants the same week. Google notes that its AI search experiences let users dig deeper with follow-up questions, which means a customer's conversation about you can go several rounds. Each round draws on the same public sources.
What this does not do
Fixing the record does not guarantee a recommendation. An assistant weighing you against a competitor with more reviews, a clearer website and more mentions will often still pick the competitor. The work improves your standing in the pool of information the AI reads, and it removes the self-inflicted losses: wrong hours, missing services, a profile nobody claimed. No one can promise a top spot in an AI answer, and any consultant who does is selling certainty that does not exist.
