A year or two ago, when someone in your town needed a family photographer, a therapist taking new clients, or an emergency plumber at 9 p.m., there was exactly one move: google it. That’s still the main road — but a new one is open, and traffic on it is growing fast. People now ask ChatGPT, Perplexity, or the AI answer at the top of Google itself: “Who’s a good wedding photographer near Nanuet?” — and get back three names with a short pitch for each.
Here’s the uncomfortable question: were you one of the three?
Most business owners have no idea, because this referral happens somewhere you can’t see. There’s no analytics dashboard for “ChatGPT mentioned you.” And that invisibility has spawned a wave of consultants selling “AI optimization” packages with the confidence of people selling something nobody can verify. So let’s do what we always try to do here: explain how to show up in ChatGPT and AI search honestly — what actually influences AI recommendations, how much of it you’re (hopefully) already doing, and where the real new work is.
By the end, you’ll know how AI assistants actually find and pick businesses, which of your existing SEO efforts count double now, and a concrete checklist for making your business easy for a machine to recommend.
The Short Answer
How do you get ChatGPT and AI search to recommend your business? AI assistants recommend businesses they can find, read, and verify on the open web. That means a fast, crawlable website that states plainly what you do and where; structured data (schema) that lets machines read your name, services, and location as facts; consistent business information across Google, directories, and review sites; and real third-party evidence — reviews, mentions, “best of” lists — that you’re a credible pick. There is no ad slot, no submission form, and no shortcut anyone can sell you. It’s the same open web Google reads, answered through a new front door.
Where AI Recommendations Actually Come From
When ChatGPT names three plumbers, that answer is assembled from two sources — and knowing the difference tells you what you can influence.
Source one: what the model already “knows.” AI models are trained on a huge snapshot of the public web. If your business has been written about, reviewed, and listed in enough places over the years, the model may have absorbed that. You can’t do much about last year’s snapshot — but everything you publish now is the material future snapshots learn from.
Source two: live web search. This is the bigger one, and the one you can act on today. Modern AI assistants don’t answer local questions from memory — they run a real web search behind the scenes, read the top results, and summarize them. ChatGPT’s search uses a live web index. Perplexity shows you its sources right in the answer. Google’s AI Overviews sit on top of ordinary Google results. In every case, the assistant is doing something familiar: reading the same pages Google ranks, and picking names out of them.
Read that again, because it’s the whole game: when an AI recommends businesses, it’s usually summarizing pages that already rank — your site, your Google Business Profile, local directories, review sites, and “best X in [town]” articles. The assistant is not a new judge. It’s a new spokesperson for the old judges.
The Good News: Most of This Is SEO You Should Already Be Doing
If you’ve been following the playbook from our guide to getting your local business found on Google, you’re further ahead than you think. The fundamentals carry straight over — and now they pay twice.
- A fast, crawlable site. AI search tools read web pages the same way Google’s crawler does. A site that’s slow, broken on mobile, or hides its content behind scripts is hard for both to read. (This is the same argument we made about Core Web Vitals and why fast sites rank higher — the audience of machines reading your site just got bigger.)
- A complete Google Business Profile. Assistants answering “near me” questions lean heavily on the sources that already organize local businesses — and Google’s own AI features draw directly on Business Profiles. Categories, hours, photos, services, and a steady flow of reviews all feed the machine.
- Real reviews, steadily earned. AI answers routinely justify a recommendation with phrases like “highly rated” or “praised for fast response times.” Those phrases are summarized from actual review text. More real reviews, with specifics in them, give the machine more to quote.
- Pages that state facts plainly. Your service area, your specialties, your pricing approach — written in plain sentences on your site, not locked in an image or implied by a stock photo.
If none of that is in place, start there — not with anything labeled “AI.”
What’s Genuinely Different About AI Answers
It’s not all the same old playbook. A few things matter more — sometimes much more — when the reader is a machine assembling an answer.
1. Question-shaped content gets quoted
An AI assistant answering “how much does a website cost?” wants a page that asks and answers exactly that. Pages built around one real question, with a clear, self-contained answer near the top, are the easiest for an assistant to lift and cite. (You’re reading that structure right now — the “Short Answer” section above exists for the same reason featured snippets do, and AI answers are, functionally, featured snippets with better manners.)
2. Structured data turns your marketing into machine-readable facts
Schema markup — LocalBusiness, Service, FAQPage, Article — is a label maker for machines. It says: this is the business name, this is the area served, these are the hours, this is a real article with a real date. When an assistant needs to state facts confidently, labeled facts win over prose it has to guess about. Schema was worth doing for Google; it’s more worth doing now.
3. Third-party mentions become the recommendation
Here’s the shift worth internalizing: when a human googles, they see your website and judge for themselves. When they ask an AI, the assistant reads ten pages about you and forms a view. That means what other sites say — the local “best photographers in Rockland County” roundup, the chamber of commerce directory, a local news mention, review sites — carries more weight than before, because it’s evidence written by someone who isn’t you. Earning those mentions is slow, unglamorous work: be listed in the directories that matter for your trade, ask happy clients for reviews, be somewhere a local writer can find you. It was worth doing before. Now it’s often the deciding vote.
4. Consistency is a trust signal for machines
If your business is “RG Marketing Group” on your site, “R.G. Marketing” on one directory, and shows two different phone numbers, a human shrugs. A machine trying to decide whether these are the same business — and whether its facts are reliable — may quietly drop you. One canonical name, address, and phone number, everywhere, character for character.
When Not to Worry About This Yet
Honesty section. If your website doesn’t show up on page one or two of ordinary Google for your own service and town, AI search is not your problem, and “AI optimization” would be money down the drain. The assistants are reading the pages that already rank; if you’re not among them, there’s nothing for the machine to find. Fix the fundamentals first — the site, the profile, the reviews. AI visibility is a second-story addition; it needs the first story built.
Also true: AI referrals are still a minority of how customers find local businesses. The share is growing quickly — OpenAI said ChatGPT passed several hundred million weekly users during 2025, and Google has rolled AI answers out across ordinary searches — but Google’s classic results and Maps still drive most local discovery today. This is a “get ahead of it” project, not a “drop everything” one. The good news, again: the work is the same work.
And a word on the tricks. You’ll run into advice about special files for AI crawlers (the llms.txt proposal), secret prompt phrases to hide in your pages, and “guaranteed AI ranking” services. Our honest read: llms.txt is a genuine proposal that costs ten minutes and can’t hurt, but as of now the major assistants haven’t committed to reading it — treat it as a cheap experiment, not a strategy. Hidden text aimed at machines is the same cloaking trick that’s gotten sites penalized for twenty years; assume it will end the same way. And any service that guarantees placement in an AI answer is guaranteeing something its seller doesn’t control. If a tactic wouldn’t make sense to do for a human reader or for Google, it almost never makes sense for the machines reading over their shoulders.
The AI Visibility Checklist
Here’s the framework we use, in order of impact:
- Pass the machine-readability test. Fast site, works on a phone, real text (not text baked into images), every service and town you serve named in plain sentences somewhere a crawler can read.
- Label your facts with schema.
LocalBusinesswith name, address, phone, hours, and area served;FAQPageon your real FAQs;Articleon your posts. If your site was custom-built, this is a small job. If your platform won’t allow it — that’s a data point about your platform. - Build question-and-answer pages for the questions you actually get. One page, one question, direct answer up top, honest detail below. Your call logs and inbox are the content calendar.
- Earn third-party evidence. A steady, FTC-clean flow of real reviews; listings in the directories that matter for your trade, with identical business info in each; any genuine local mention you can earn.
- Check your own visibility quarterly. Ask ChatGPT, Perplexity, and Google’s AI the questions your customers ask — “best [your service] near [your town],” “should I hire X or do Y myself” — and see who gets named. It’s the closest thing to an analytics dashboard this channel has, and it tells you exactly which third-party sources the assistants are leaning on in your area. Then go get mentioned there.
What We’d Tell You Over Coffee
The AI-search gold rush is producing a lot of confident selling and very few verifiable claims. Nobody — including us — can promise you a ChatGPT recommendation, and anyone who does is telling you something about themselves. What we can say from the mechanics: assistants recommend businesses that are easy to read, easy to verify, and vouched for by someone other than themselves. That’s not a trick. It’s the same definition of a credible business the web has always rewarded — enforced now by a reader that never skims.
If your site gets traffic but no calls, or you’re not sure whether a machine could even read your current site, that’s exactly the kind of thing we look at in a site review. Get in touch — or if you’re weighing a rebuild, our pricing breakdown shows what honest numbers look like. Either way: the next customer who asks a chatbot for a recommendation has already decided to hire someone. The only question is whose name comes back.
