Where AI assistants actually get local business data

Ask ChatGPT for a plumber in Clearwater and it does not run a Google search on your behalf. It consults places data it licenses — Foursquare first — and confirms details against the Bing index. If a business is absent from that data, or present with a stale category, a dead phone number, or an old address, the assistant either skips it or repeats the error with complete confidence.

About the share, since this page is where the number lives: the widely quoted "60–70% from Foursquare" traces to one 2025 test by the Spanish consultant Natzir Turrado, who read ChatGPT's internal places payload across fifty prompts in five Spanish cities. That is the only attributable measurement of it anyone has published. It is directionally useful and far too small a sample to state as a rate for the United States, so this page names the mechanism as dominant and leaves the percentage to its author.

Foursquare's consumer app faded years ago, which is exactly why its listings sit unclaimed — and why the data behind the most-used assistant is still cheap to fix. The listing costs nothing, and the businesses that get there first inherit a mostly empty field.

Nothing on that list is exotic. These are ordinary business listings, claimable in an afternoon at no cost. The point is the shape of the coverage map: five rows, and the typical local business has filled exactly one.

Budget patience after the claim: corrected listing data generally takes two to six weeks to start surfacing in assistant answers, and content and entity work compounds over months, not days. Claim now precisely because the payoff lags.

The scale of that gap is documented: SOCi's 2026 index, built on more than 350,000 locations, found ChatGPT recommending a little over one local business in a hundred while Google's local pack covers more than a third of them.

Read the gap as scarcity, not irrelevance. An assistant can only recommend a business it can find in the data it reads and whose facts agree — so being among the few it can see is worth a great deal, and the path in runs through listings your competitors have not claimed.

None of this argues for abandoning Google Business Profile. The local pack remains the largest single recommendation surface, and the profile deserves ongoing work. The mistake is exclusivity: treating the one listing Google reads as if it were the whole data layer.

Published studies put that conversion multiple in the high single to low double digits over organic, and the funnel splits behind it are compiled with their sources on our statistics page. Someone who asks an assistant to name a provider is not starting their research. They are ending it, and they tend to contact whoever gets named.

That combination — demand up several-fold in a year, conversion at multiples of organic, and a data layer nobody maintains — is the entire case for acting before the window closes.

The work is also verifiable. You do not have to take anyone's word for it: open each platform and look at your own listing. If it is unclaimed, wrong, or missing, that is what every assistant reading it sees too.

One honest caveat. Claiming listings makes a business retrievable; it does not guarantee any assistant will name it, and nobody honest promises that. What can be committed to is clean data at every source and visibility measured as a trend across many prompts — the structure of Hitman Marketing's work, from the $497 AI Visibility Teardown, which reports claim status across all of these platforms, to search retainers from $895/mo that include claiming and normalizing them in every tier.