AI search now recommends local businesses by name, and most businesses never show up. Only 1.2% of locations get recommended by ChatGPT, 11% by Gemini and 7.4% by Perplexity, against 35.9% for Google’s traditional local 3-pack (SOCi, 2026). Two components decide whether you appear: a map fed by structured business data, and text fed by web pages and citations. Fix one and the other stays broken. Most consultants selling « AI SEO » packages are treating this as one problem.
It is two, and they run on different fuel.
The Terminator Was Never Going to Stop
Terminator is not about killer robots. It is about a system that does not get tired, does not take a day off, and does not care whether you noticed it running. That is what AI search discovery has become for local business. It runs continuously across ChatGPT, Gemini, Perplexity and AI Overviews, pulling from your Google Business Profile and your website whether you touch it or not. You can either own the infrastructure that feeds it, or you rent your visibility from whoever else’s page it decides to cite instead.
The businesses winning right now are the ones that built the machine before they needed it.
AI Search Is Already Part of the Buying Journey
AI-driven local discovery is no longer experimental. It is the front door for a growing share of customers before they ever see your website.
28% of people have visited a local business they had never heard of purely because an AI recommended it, based on Yext’s survey of 3,848 consumers in 2026. More than four in ten people now use an AI tool to search for a local business in any given month, and 71% of those say they are doing it more than they did a year ago.
Consumer adoption has moved fast. 45% of consumers now use AI tools to find local services, up from 6% a year earlier (GrowthPro AI, 2026). Despite that shift, 88% of local businesses have no active strategy for AI search, according to the same research.
That gap is the opportunity. AI Overviews now appear in 68% of local searches, and when one shows up, click-through rate to the actual business drops from roughly 15% to 8%. Fewer clicks, more concentration. The top 20% of businesses now capture 68% of local search visibility, up from 52% in 2023.
Being named in the answer, not just listed somewhere below it, is the objective. Everything else is a rounding error.
Two Pipes, Two Fixes
A local AI answer is built from two separate systems, and optimizing one does nothing for the other.
The map component is drawn from structured business records: name, address, hours, rating, phone. On ChatGPT in the UK, that map is filled directly from Google My Business, confirmed by tracking parameters in the outbound link that match the exact URL sitting in a business’s Google Business Profile. In the US, the picture splits: most results still link to the business’s own site, but a meaningful share route through Yelp instead.
The text underneath the map comes from an entirely different source: web pages, with the citations shown as proof. A business can be a pin on the map and invisible in the text, or cited five times in the text and absent from the map completely.
This is why a business can spend a year on backlinks and content and still not appear when someone asks « best gym near me. » The map runs on structured data. The text runs on crawled pages. Neither substitutes for the other.
Your Google Business Profile Is Doing More Work Than You Think
A complete Google Business Profile is the single highest-leverage fix available, and most businesses treat it as a one-time setup task instead of a living asset.
On Google’s own AI surfaces, the Google Business Profile itself is the most cited source, not the business website. It accounts for 93% of AI Mode’s top citations and 65% of AI Overviews citations. Everywhere else, from Gemini to ChatGPT, the business’s own website and social profiles do the heavy lifting.
Six elements of a Google Business Profile feed directly into AI answers, and each one gets checked separately by the models pulling from it.
Categories need the narrowest accurate primary category plus genuine secondaries, because Google changed its available category list roughly 40 times last year. Attributes such as accessibility and identity checkboxes now determine whether you appear for qualified searches like « wheelchair accessible gym near me. » Services need names, descriptions and price ranges in text a model can actually parse, because unpriced services do not get named when someone asks what something costs. Photos are read by Google to infer what you offer, so stale image libraries quietly suppress you. Descriptions should use the full 750 characters available, and hours need to be accurate including holidays. A real local phone number, not a generic national line, functions as a geographic trust signal.
41.3% of people check a Google Business Profile after receiving an AI recommendation. A wrong phone number does not cost you a search ranking. It costs you the customer at the final step, after the AI already did its job.
Citations Are Not the Same Business Twice
The sources an AI cites for the same brand change completely depending on which platform someone asks.
Tracked across one independent multi-site gym brand, the business’s own website was the single most cited domain on Gemini, appearing nearly double its nearest competitor. On ChatGPT, that same website ranked fifth, behind three separate aggregators and directories. On Perplexity, the business’s own site did not appear in the top five sources at all.
That means a business can dominate one AI platform and be structurally invisible on another, using identical content. There is no single citation strategy that covers all of them. Facebook and Instagram outrank most individual business websites as citation sources on AI Mode and AI Overviews specifically, which means social profiles are not a marketing afterthought anymore. They are a ranking input.
Location pages carry more weight than homepages inside this system. Tracked across 3,445 cited pages in one category, a brand’s Welwyn Garden City location page was cited across 45 separate prompts, its Stevenage page across 40. Competing national chains showed the identical pattern on their own town-specific pages. No page for the town means no presence in the answer, regardless of how strong the homepage is.
Presence Beats Everything, Including a Head Start
A location without a physical address gets zero visibility, no matter how strong the brand is everywhere else.
In one tracked case, a gym brand with two dominant established locations scored zero across every tracked prompt in a nearby affluent town where it had no physical premises. No Google Business Profile, no location page, no presence in the map or the text, despite being eleven miles from a location that scored 90% visibility on some queries. Two national chains with physical sites in that town owned every answer instead. The AI was not overlooking the absent brand. It had nothing to retrieve.
Here is the part most local operators miss: presence can start before the doors open. A location due to launch in early 2027 already showed up in 47% of tracked answers for wellness and recovery facilities, and 46% for a specific named amenity, purely because the page existed, the town was named and the offer was described in advance. A gym nobody could join yet was already competing with gyms that had operated in that town for years.
Presence is the price of entry. Once you have a real address with a fully attached profile and location page, you can outrank national chains in your own town. Skip it, and no amount of content elsewhere saves you.
Comparison: What Feeds Each AI Search Surface
| Signal | Google AI Mode / Overviews | ChatGPT | Gemini | Perplexity |
|---|---|---|---|---|
| Primary citation source | Google Business Profile (93% / 65%) | Business website, aggregators (e.g. category directories) | Business’s own website (dominant) | Third-party directories, competitor sites |
| Map data origin | Google My Business | Google My Business (via GBP link) | Google My Business | Mixed, less map-dependent |
| Social profile weight | High (Facebook, Instagram rank above most websites) | Moderate | Low | Low |
| Location page value | High | High | High | Moderate |
| Fastest fix | Complete GBP: categories, attributes, priced services | Priced service pages, aggregator listings | On-site location pages | Third-party directory presence |
The pattern across every column: no single fix covers every platform, but a complete Google Business Profile and a real location page cover the most ground on the most surfaces.
Two Technical Fixes Nobody Checks
A canonical tag is a hint, not an instruction, and AI crawlers treat it accordingly.
One brand was cited as example.co.uk on one platform and www.example.co.uk on another, for identical content. The canonical tag was correctly set. It was ignored anyway, because both versions of the URL returned a live page instead of one redirecting to the other. Anything crawling the site could reasonably treat them as two separate places, splitting citation weight across both. The fix is a server-level 301 redirect from one version to the other, applied the same way to trailing slashes, http versus https, and tracking parameters.
NAP consistency, name, address and phone identical everywhere, still matters as much in 2026 as it did before AI search existed. It is the difference between a model retrieving one clean business record and retrieving three conflicting ones, and choosing not to use any of them.
The Verdict
Most local businesses are optimizing for a search engine that stopped being the only front door two years ago, while 88% run zero strategy for the system actually deciding who gets named. The fix is not complicated: a fully attributed Google Business Profile, a real page for every town you serve with prices a model can read, consistent NAP down to the trailing slash, and location pages that name what you actually have before a competitor’s page does it for you. None of that requires a redesign or a content team. It requires someone checking it every few months, the way you would check inventory or payroll, because the system does not stop checking either. Businesses that treat this as infrastructure they own, not a listing they filled in once, are the ones showing up before their doors even open. That same own-versus-rent logic applies past local search. Renting your AI visibility means depending on someone else’s platform, someone else’s crawl priority, someone else’s decision to cite you or not. Owning the infrastructure behind your AI presence, the way Asymmetriq approaches AI systems for businesses that want control instead of dependency, costs roughly $4,700 a month less than most people assume renting equivalent capability does over time.
FAQ
Q: Do I need a different strategy for ChatGPT versus Google AI Overviews? A: Yes. Google AI Overviews and AI Mode pull most citations from your Google Business Profile, while ChatGPT leans more on your own website and third-party aggregators. A strategy built only around Google Business Profile will underperform on ChatGPT, and vice versa.
Q: Is it worth building a location page before a new site opens? A: Yes. Data from one tracked gym brand showed a location due to open in 2027 already appearing in close to half of relevant AI answers, purely because the page existed and named the town and the offer in advance.
Q: Why does my business show up on one AI platform but not another? A: Each platform weights citation sources differently. One brand’s own website was the top-cited source on Gemini and fifth on ChatGPT, absent from Perplexity’s top five entirely, using identical content across all three.
Q: Is fixing my Google Business Profile actually worth it, or is that outdated advice? A: It is not outdated. On Google’s own AI surfaces, the Business Profile itself, not the website, is the single most cited source behind the answer, accounting for the large majority of top citations in AI Mode and AI Overviews.
Q: What’s the fastest way to find out if my business is invisible in AI search? A: Run ten prompts your actual customers would use, across ChatGPT, Gemini and AI Mode, once per town you serve. Note who gets named and in what order, then click through the citations to see exactly which pages the AI is pulling from.
Q: Why do most local businesses get this wrong? A: They treat AI search as an extension of traditional SEO and optimize content alone, while ignoring that the map component runs entirely on structured business data that content never touches.
Q: Does having no physical location in a town mean I have zero chance there? A: Functionally, yes. One tracked brand scored zero across every prompt in a town eleven miles from a location scoring 90% visibility, because there was no address, no profile and no location page for AI to retrieve. <!– JSON-LD schema: Article + FAQPage to be added at CMS publish step –>