How to get found in ChatGPT and AI Overviews.

Open ChatGPT right now and type: best [your service] in [your city].

You’ll get an answer. Not ten blue links to evaluate. Not a map with three options. One answer, maybe two or three names, delivered with total confidence as if it were simply the truth.

Is your business one of them?

For most local businesses reading this, the answer is no. And here is what makes it dangerous: there is no notification. No ranking-drop alert. No line in your analytics that says a customer considered your industry, got handed three names, called one of them, and never knew you existed. The loss is completely silent.

This is the fastest-moving shift in how customers find local businesses. So let’s go through it properly: what the research actually says, what it means for a small business, and which tactics have evidence behind them versus which are being sold hard right now with nothing to back them up.

The adoption numbers

Local marketing shifts usually creep. This one sprinted.

BrightLocal’s 2026 Local Consumer Review Survey (1,002 US consumers, published February 2026) found that 45% had used AI tools to find a local business in the past year, up from 6% a year earlier. AI is now the third most used tool for local business recommendations, behind only Google and Facebook, and ahead of Yelp and TripAdvisor. ChatGPT leads at 31% of consumers, Google’s AI Mode at 23%, Gemini third.

The platform numbers line up. OpenAI said ChatGPT passed 900 million weekly users in February 2026. Google said at I/O in May that AI Mode had passed a billion monthly users, with queries more than doubling every quarter. Pew Research Center puts the share of US adults who use AI chatbots at 49%, up from 33% in 2024.

Meanwhile the ordinary results page keeps shrinking as a source of clicks. SparkToro’s June 2026 analysis of Similarweb clickstream data put the US zero-click rate at 68%, up from roughly 45% a decade ago. And Pew’s 2025 behavioral study of 68,879 real Google searches found people clicked a traditional result 8% of the time when an AI summary was present, against 15% when it was not.

Now the number that gets quoted most in this conversation, which deserves better handling than it usually gets. Whitespark’s 2025 study of 540 queries across three cities and six industries found AI Overviews on 68% of local business queries, against 39% for the local pack. That average hides an inverse relationship worth knowing about: for plain local-intent searches — someone looking for a business to call — the local pack appeared 93% of the time and an AI Overview only 15%. For informational queries it flipped, with AI Overviews on 92%.

So the Map Pack is not going anywhere for “plumber near me.” What AI has taken over is the research question that comes before it, and the whole conversation when someone opens ChatGPT instead of Google in the first place. Both matter, and they are won in different places.

Your Google rankings don’t automatically transfer

Here is the finding that should reframe the rest of this.

SOCi’s 2026 Local Visibility Index, covering more than 350,000 locations across 2,751 multi-location brands, found only a 45% overlap in retail between the brands most visible in traditional local search and the ones AI platforms recommend most. You can be in the Map Pack, ranking well and collecting calls from Google, and still be absent when someone asks ChatGPT the same question.

Then the supply side, which is where the opportunity sits. In that same index, ChatGPT recommended just 1.2% of the locations tested, against a 35.9% average appearance rate in Google’s local three-pack. Demand is mainstream. Supply is nearly empty.

One honest caveat, because this post is about accuracy rather than urgency. AI referrals still average about 1% of total website traffic, measured by Conductor across 13,770 domains and 3.3 billion sessions, even though ChatGPT’s outbound referrals alone grew 206% between January 2025 and January 2026. Don’t abandon local SEO for this. Treat AI visibility as the fastest-growing layer on top of the foundation you already need.

The quality of that small slice does look different, with the usual caveats about who paid for the research. Semrush’s mid-2025 study found AI search visitors converting at 4.4 times the rate of organic search visitors — in digital marketing topics, the vertical with the highest AI adoption, and without a published sample. Adobe Analytics, across more than a trillion visits to US retail sites through July 2025, measured AI referral visits running 32% longer with a 27% lower bounce rate. Someone arriving from an AI recommendation has been handed your name as the answer.

What the evidence says doesn’t work

Before the things that do work, let’s clear out what is being sold hardest. If an agency pitches you on AI visibility and leads with either of the first two, you now know more about this than they do.

On schema specifically, that same Ahrefs study found AI-cited pages were almost three times more likely to already carry JSON-LD — exactly the kind of correlation that gets misread as causation. Ahrefs attributed it to schema living on better-maintained sites that also publish stronger content and earn more links, while stopping short of ruling out a real effect.

Two caveats in the other direction, since this post is about being accurate rather than breathless. Every page in that study already carried at least a hundred AI Overview citations, so it tests whether schema lifts pages that are already visible, not whether it helps an unknown business. And the extraction test covers what these systems read when they fetch a page in the moment; schema is still very likely parsed during ordinary indexing. Keep your schema. It supports rich results, crawl clarity and plain disambiguation about who you are. Just stop treating it as an AI visibility lever.

Google’s own Search Central documentation says a page needs only to be “indexed and eligible to be shown in Google Search with a snippet,” and that “you don’t need to create new machine readable files, AI text files, or markup… There’s also no special schema.org structured data that you need to add.” Google’s John Mueller put the file question more bluntly in June 2025: “FWIW no AI system currently uses llms.txt.”

For the record, this site publishes an llms.txt. It took ten minutes, it costs nothing, and we would never put it on an invoice.

What the evidence says does work

Now the useful part, ordered roughly by impact for a local service business.

1. Your Google Business Profile is the spine of local AI answers

When ChatGPT, Gemini or Google’s AI Overviews put together a local recommendation, your Google Business Profile is a primary source. Categories, services, hours, location, attributes, photos, reviews: that is the structured, verified information these systems lean on when deciding who exists and who is worth naming.

A thin or half-finished profile doesn’t only hurt your Map Pack position. It gives AI systems almost nothing to work with. If your services section is empty, your categories are wrong and your last photo went up three years ago, you have made yourself hard to recommend.

That is good news for small businesses, because the highest-impact AI visibility work is the same work that improves your Google rankings. One effort, two channels.

2. An active review profile

This is the largest effect in the post, and it comes with an asterisk worth reading.

Seer Interactive, working with Trustpilot, looked at 804,491 AI responses to 15,783 prompts covering 1,926 brands across ChatGPT, Gemini, Perplexity and Google AI Mode. Brands with no active review profile were cited in a median 1% of answers. Brands with a profile reached 53.5%. Brands with roughly eighty or more reviews and regular responses reached 75%.

The asterisk: Trustpilot commissioned the study and the tiers are defined by Trustpilot profiles, so read it as strong evidence about review presence rather than a law of physics. Two findings in the same data are harder to argue with. Review and trust sites made up 14% of all citations, second only to brand websites — and 24% of citations at the point where someone is deciding who to buy from.

The consumer side points the same way. BrightLocal found 97% of consumers read reviews for local businesses and 41% now always do, up from 29%. And 88% of the people using AI for recommendations fact-check what it tells them, about half by checking whether a cited review is genuine. Reviews do double duty: getting you into the answer, then convincing the person who reads it.

If you take one action item from this post, take this one. Build a consistent system for asking for reviews and responding to every one of them.

3. Off-page mentions, because most citations aren’t from your site

Omniscient Digital analyzed 23,387 citations across 240 branded prompts and found owned brand content accounted for just 23% of them. The other 77% came from earned media and third-party commercial content: directories, review platforms, forums, video, news.

That reframes the job. You can have a perfect website and still be invisible, because most of what these systems read about you was written by other people, somewhere else. For a local business that means directory consistency across Bing Places, Apple Maps, Yelp, Foursquare, the BBB and your trade’s own directories. It means local news mentions, chamber listings, supplier and partner pages, and community involvement that ends up in writing somewhere. It means being discussed, not merely present. (240 prompts is a modest base, so take the proportion as a strong signal rather than a precise constant.)

4. Content an AI can actually extract

Because these systems read what is actually in your HTML, how the page is built matters. Testing more than a hundred ChatGPT Agent sessions for Search Engine Land in October 2025, Jes Scholz found 46% of visits began in reading mode — a text-only view that strips the page back to raw HTML and links, with no images, CSS, JavaScript or schema.

Machines are a real share of the audience now. Cloudflare reported in June 2026 that automated requests had passed human ones for the first time, at roughly 57% of HTML requests, a crossover the company had not expected until 2027.

Practically, that means content locked behind JavaScript, hidden in tabs or accordions that don’t render server-side, or baked into an image is content an AI cannot use. And when ChatGPT can’t read your page, it reads third-party pages about you instead — which is how a business ends up with old pricing repeated confidently in an AI answer.

The fix is unglamorous and effective: real text on the page, in clean HTML, that loads fast.

5. Specifics, not adjectives

The original research here is a Princeton-led paper, presented at KDD 2024 and benchmarked on 10,000 queries. Measured against an unoptimized baseline, adding quotations lifted visibility 41% on its main metric, adding statistics 30%, and citing sources 28%. Keyword stuffing made things worse, at minus 9%.

Treat those exact numbers as experimental — they were measured inside a Bing-Chat-style system in 2023 and 2024, not today’s AI Overviews — and the paper itself notes results vary by industry. The direction, though, matches everything else here: these systems favor content that is specific, verifiable and attributable.

For a local business, that means writing like an expert instead of a brochure. Real numbers, real process, real specifics. “We pressure test every pontoon compartment and every fuel tank” is extractable and quotable — that line comes straight from a client’s own answers to customer questions. “Quality craftsmanship you can trust” is not.

Question-shaped headings help too, because that is how people talk to a chatbot. Build pages around the questions customers actually ask, and answer each one directly in the first sentence underneath it.

Test your own AI visibility in ten minutes

You don’t need a tool or a subscription to find out where you stand.

1. Open ChatGPT, Gemini and Google’s AI Mode. In each, ask for the best [your service] in [your city], phrased the way a customer would say it rather than in industry terms.
2. Write down every business named. Note whether you appear, in what position, and how you are described.
3. Ask a follow-up the way a real customer would: who is most affordable, who handles emergencies, who is best for [specific need]. Narrowing questions often produce different names.
4. Ask each tool directly what it knows about your business by name. This is how you find out whether AI systems are repeating old hours, a wrong number, or services you dropped.
5. Run a few Google searches for your services and note how often an AI answer appears above the results, and whether anyone from your market is cited in it.
6. Check your own site the way a bot sees it: turn off JavaScript, or use your browser’s reading mode, and see how much of your content survives.

If you don’t appear anywhere in those tests, you are in ordinary company — in SOCi’s index, ChatGPT named barely one location in a hundred. If you appear with wrong information, that is more urgent than not appearing at all, because confident misinformation costs you calls.

Where this is heading

Two trends worth watching, because they change the calculation over the next year.

First, AI referrals are fragmenting. In Goodie’s panel of 41 brand sites and 2.8 million AI referral sessions, ChatGPT’s share fell from 89% to 63% over eight months, while Claude climbed from 1.4% to 18.5%, Gemini quadrupled and Perplexity more than doubled. Other panels put ChatGPT much higher, so trust the direction rather than the levels. Either way, optimizing for one assistant is already the wrong plan, and the fundamentals travel across all of them.

Second, citations are drifting away from the rankings. Ahrefs, across 863,000 keywords and four million AI Overview URLs, found citations drawn from top-10 organic pages fell from 76% in July 2025 to 38% in March 2026. BrightEdge, on a different dataset, puts it nearer 17%. The methods disagree; the direction doesn’t.

That second one is encouraging for a small business. If citations came strictly from the top ten, competitors with decade-old domains would own AI visibility permanently. That is not what is happening.

AI search FAQ

Do I need a special AI SEO service to appear in ChatGPT?

Not a separate one. The work that drives AI visibility for a local business is largely the work that drives local SEO: a complete, active Google Business Profile, a steady flow of reviews with responses, consistent directory listings, and specific, fast-loading content. Be skeptical of anyone selling AI visibility as a wholly separate product, especially when the deliverables are schema markup and an llms.txt file.

Does ranking #1 on Google mean ChatGPT will recommend me?

No. SOCi’s 2026 index of more than 350,000 locations found only a 45% overlap in retail between the brands most visible in traditional local search and the ones AI platforms recommend most. They are separate systems with different inputs.

Is schema markup useless now?

Not useless, just not an AI visibility lever. Ahrefs’ controlled study of 1,885 pages found no meaningful citation lift from adding JSON-LD, and AI Overviews citations actually fell slightly, while a separate live-fetch test found the major systems reading only raw HTML. Schema still supports rich results, crawl clarity and entity disambiguation, so keep it. Just don’t pay extra for it as an AI service.

Should I add an llms.txt file to my site?

There is no evidence it helps. Ahrefs checked 137,210 domains and found 97% of valid llms.txt files got zero requests in May 2026, with AI bots never going looking for one, and Google has said plainly that no AI system uses them. We publish one on this site anyway, because it costs nothing. It should not appear on an invoice.

How much traffic will AI search actually send me?

Today, roughly 1% of your total traffic — Conductor measured 1.08% across 13,770 domains and 3.3 billion sessions — though ChatGPT’s outbound referrals alone grew 206% year over year. The value is in the quality rather than the volume: Semrush measured AI search visitors converting at 4.4 times the rate of organic visitors in its own vertical, and Adobe measured 32% longer visits with 27% lower bounce rates.

What is the single fastest way to improve my AI visibility?

Reviews, then your Google Business Profile. In the Seer and Trustpilot study of 804,491 AI responses, brands with no active review profile were cited in about 1% of answers, against 75% for brands with a real review presence and regular responses. Those two moves cover most of the available upside for a local business.

Where we fit

We built this agency around AI, so we have been watching this shift closely — and we are not going to sell you a separate AI package with a text file in it.

The work that gets a local business found in ChatGPT and AI Overviews is the work already inside our Growth plan: Google Business Profile optimization maintained continuously rather than configured once, Google Business posts three times a week, review generation with monitored responses, citation consistency and listing monitoring, local landing pages and ongoing content written with real specifics, and fast, clean sites that bots can actually read.

None of that was built for AI search. It turns out that doing local SEO properly is what AI systems reward, which is why we have not had to invent a new product to address this. We have had to sharpen an existing one.