Nearly half of people now ask an AI to find a local business. The AI names almost nobody.

Somewhere this week, somebody who needed exactly what you sell opened ChatGPT instead of Google. They typed something like "who does end-of-lease cleaning near me and what should it cost," and got back an answer with three businesses named in it. Not ten blue links to work through. Three names, in a sentence, with a reason attached to each. If yours wasn't one of them, you never found out the enquiry existed.

Twelve months ago you could file that under someday. You can't now. BrightLocal's 2026 consumer survey found 45% of people had used an AI tool to find a local business in the past year — up from 6% the year before. ChatGPT on its own accounted for 31%. Over the same stretch, the share of people using Google Reviews fell from 83% to 71%. It's a US survey of about a thousand adults, so hold the decimal places loosely. A seven-fold jump in a single year isn't a rounding error, though, and Australia has never lagged the US on this stuff by more than a year.

The bit that should actually worry you

Here's the other half of the picture, and it's the half nobody puts in the newsletter. SOCi's 2026 Local Visibility Index ran the numbers on nearly 350,000 business locations across a couple of thousand brands, and checked how often each one actually got named by the AI assistants. ChatGPT recommended 1.2% of them. Perplexity managed 7.4%, Gemini 11%. The same locations turned up in Google's local 3-pack 35.9% of the time.

Sit those two facts next to each other. The number of people asking an AI to pick a business for them is climbing fast, and the number of businesses the AI is willing to pick is tiny. That's not a gap you close by trying harder at the thing you were already doing, because the two systems aren't doing the same job.

A ranked list is forgiving. Ten results on a page, a map pack, a second page behind it — being eighth is a bad day, not a death. Some people scroll. An answer isn't a list. When an assistant writes a sentence naming three businesses, everyone else in your suburb was not ranked lower. They were left out of the sentence.

Winning at Google no longer means being found

The uncomfortable finding in that same index: fewer than half the brands leading on Google local visibility also showed up among the leaders in AI results. In retail the overlap was 45%. Which means more than half the businesses winning the map pack in their category are simply absent from the AI answers about that same category.

That surprises people who've spent years and real money on local SEO, and it shouldn't. The two systems are solving different problems. Google's job is to hand you a set of plausible options and let you sort them out — it can afford to include a business it's only half sure about, because you'll click through and judge. An assistant writing prose has to commit. It's putting its own credibility behind a name. So it doesn't ask "who ranks well here." It asks a much harsher question: which of these can I confirm well enough to say out loud?

The filter isn't judging your marketing. It's judging whether it can corroborate you.

That reframing is the whole practical point of this post. The businesses getting named aren't the ones with the cleverest copy. They're the ones whose basic facts line up everywhere the model looks — the same trading name, the same phone number, the same suburbs listed, the same services, on your own site and on Google Business Profile and on Facebook and on whatever directory you signed up to in 2019 and forgot. Every place your details disagree with each other is a reason for a filter to skip you and name someone less ambiguous.

Ratings do the same job. In that index, the locations that got recommended averaged somewhere between 4.1 and 4.3 stars depending on the platform. Not five. Not a hundred reviews. Just credibly above the line, from real people, in more than one place. The filter is looking for corroboration, not brilliance.

And here's the genuinely encouraging part, which almost nobody says out loud: those 350,000 locations were multi-location brands. Companies with marketing departments and agencies on retainer. They're the ones scoring 1.2%. A chain with forty branches has forty chances for its hours, its phone numbers and its service lists to drift out of sync. You have one. On the one axis that actually decides this — is this business consistent and confirmable — a careful sole operator can beat a national brand, and frequently does.

• Make your details identical everywhere, down to the punctuation. Trading name, address, phone, hours, and the suburbs you actually service. Contradictions are the single cheapest reason to be filtered out.
• Say plainly what you do and where, in the words a customer would use. "Office and end-of-lease cleaning, and here are the eleven suburbs we cover" is confirmable. "Excellence in integrated facility solutions" is not — a model can't tell what you do or where you do it from that, so it won't risk saying anything.
• Get reviews to credibly above-average and keep them coming. The recommended businesses sat around 4.1–4.3 stars, so this is a threshold to clear, not a leaderboard to top.
• Answer the boring questions in public — what's included, rough price ranges, after-hours, insurance, ABN. That's the material an assistant is trying to summarise when someone asks. If it isn't written down anywhere, you're asking it to guess, and it would rather name somebody else.
• Then check. Once a month, ask ChatGPT, Gemini and Google's AI mode what they'd recommend in your suburb for what you do. It's free, it takes five minutes, and most owners have never done it once.

Don't buy the acronym

There is already an industry forming around this, with fresh three-letter names for it, and invoices to match. Be careful there. Read the list above again — it's business-detail hygiene, honest public answers, and real reviews. That's the same unglamorous work that has always made a small business findable. It has been repackaged, not replaced.

Which isn't to say nobody can help you do it; plenty can, and if listings cleanup is a job you'll never get to, paying someone is a fine call. Just know what you're paying for. If a pitch can't explain in plain words which of your facts are currently inconsistent and where, it's selling you the acronym, not the work.

Where we land on it

Most of what's above isn't something we sell, and we're not about to pretend otherwise — nothing in our product will tidy up your Google Business Profile for you. What we do have a stake in is the half that starts after the AI names you. Because the shift changes that half too: when someone arrives having already been told you're the right business, they arrive further down the track and less patient. Inside Dispatch, the AI reads the enquiry and drafts the reply — quoting the standard jobs off your own price list — so the 9:40pm one gets a real answer at 9:41 instead of 8am tomorrow. The send button still sits behind you.

That's the honest shape of it. Being named by the assistant is won months earlier, in the dullest possible work: your own details, straight, everywhere, backed by real reviews. Nobody sells a course on that because it doesn't need one. But the mechanism underneath has genuinely changed, and it's worth understanding what changed — search used to rank you, and you could live with mediocre. It filters you now, and filters don't do mediocre. They do in, or out.