When ChatGPT tells someone the three best coffee shops in their neighborhood, it isn't remembering a visit. It's putting an answer together from sources it has learned to trust. Once you know which sources those are, you can stop hoping you get recommended and start seeing why you don't.
In this article I'll go through the platforms AI assistants seem to lean on most for local recommendations, and what being present on each one actually takes.
Why do sources matter so much?
An AI assistant answering a local question has a hard job. It can't check anything in person, so it relies on sources that a lot of people have already contributed to and that carry signs of trust: number of reviews, ratings, how recent they are, and whether people agree. When those sources agree that a business is a strong option, the model repeats it. When a business is missing from them, the model has little to draw on and often leaves it out.
And the assistants don't all read the same way.
of the businesses AI named in our 48-market study were named by only one of the four assistants.
That was the result that surprised me most. If the four assistants were reading the same short list of sources in the same way, you'd expect a lot more overlap. What it suggests to me is that each assistant weighs sources differently, so being strong on one platform may get you named by one assistant and skipped by the others. That's one of the reasons I'd try to be present across several of the sources below instead of betting on one.
If you haven't read my overview of what AI visibility is, it's a good companion to this piece.
The usual suspects
Google Business Profile and Maps
This is where most local trust starts. Your Google Business Profile feeds Maps, feeds local search, and feeds the facts assistants repeat about you: category, hours, location, rating and review count. An incomplete or unclaimed profile is one of the most common reasons a business looks weak to a model, and one of the easiest to fix.
What being present means here: claim the profile, fill every field, keep hours and category accurate, add photos, and keep earning reviews. When we scan a business that never appears in AI answers, a thin or unclaimed Google profile is often part of the story. The free profile check is a quick way to see where yours stands.
Review platforms
Large review sites carry the signals AI trusts: many independent voices, star ratings and recent activity. Depending on your category that might mean the general platforms, or a platform specific to your industry. A competitor with a few hundred reviews on the right platform isn't necessarily better than you, but the model has a lot more to read about them, and that's what gets repeated.
What being present means here: claim your listings, keep the information consistent with your Google profile, reply to reviews and ask happy customers to leave one. Both the number of reviews and how recent they are matter. If you want to know how far behind a competitor you are, you can check that with the review gap tool.
Community discussion
Threads where real people ask "who should I use for X in this city?" and get named answers can be very useful to an AI assistant, because they look like the kind of honest recommendation it wants to pass on. These conversations happen on community platforms and local forums, and they may carry more weight than their size would suggest.
What being present means here: you can't fake this, and you shouldn't try. What you can do is do work worth talking about, join in honestly where your business is being discussed, and never post fake endorsements.
Curated "best of" lists and local guides
"Best gyms in the city" articles, local publication roundups and niche directories work like ready-made shortlists. When several of them name the same business, an assistant is likely to treat that as strong confirmation.
What being present means here: get onto legitimate local lists through outreach, awards and being the kind of business editors want to include. Keep an eye on the roundups that rank for your category, because those are often the pages AI is reading.
Your own site and its structured data
Your website usually isn't the main source AI trusts, but it's where the model can confirm facts, and structured data makes those facts clear. Schema markup tells a machine, in a format built for machines, your name, address, category, hours and rating.
What being present means here: publish clear, factual pages and add LocalBusiness schema so the details are explicit. The LocalBusiness schema guide explains how, and the free schema generator creates valid markup you can paste in.
How do you find your own sources?
The most useful exercise is also the simplest. Ask an assistant the question your customers would ask, then read the answer closely and note which platforms it mentions or clearly relied on. Do it a few times, because answers vary between runs. Patterns show up quickly, and you'll usually see the same few sources doing most of the work in your category. If you're not sure what to ask, the prompt generator helps you write questions the way a customer would.
Once you know your sources, the work gets concrete. For each platform doing the heavy lifting, ask yourself: Am I on this platform, and is my information correct? Which competitor does the AI keep naming, and how many reviews do they have there compared to me? A competitor the AI names again and again is usually strong on a source where you're missing or thin, and that difference is your to-do list.
The free quick check tool shows you whether you appear for a "best in your city" question and surfaces the sources behind the answer, which is a fast way to start this exercise.
Keeping it up
Being present on these sources isn't a one-time project. Reviews pile up and go stale, competitors claim listings, new roundups get published, and AI answers shift as all of that changes. I'd check your sources the way you check your books, regularly rather than once. For the step-by-step version, read how to get recommended by ChatGPT, which turns the sources above into an ordered plan.
LocalSeen.ai scans the AI answers in your category on a schedule, shows which sources they cite and marks whether you and each competitor are present on them, so if you want to see your own source map, start a scan.