We measure who AI recommends. It keeps getting local wrong.
People now ask AI assistants to recommend a local business the way they used to ask Google. So we ask the assistants the same questions real customers do, count every business they name, verify each one against the live web, and publish the result. More than 500 answers across six New York markets measured so far, three published below and more added monthly. The pattern is consistent, and it is not flattering to AI.
ChatGPT + Claude + Gemini, verified names only, method published, updated monthly, CC BY 4.0
How AI gets local wrong
Different market, same cause: if your online signals are thin, the assistant cannot find the real you, so it reaches for a chain, a bigger advertiser, or a stale name the web still remembers.
The July 2026 sample asked the assistants directly, without web browsing or location, to see what they knew on their own. Refreshes from August 2026 on sample with live web search, the way a customer's assistant actually answers. What does not change: if the web does not describe you clearly, neither mode can recommend you.
Movers, NYC: a chain swallows the market
One national brand was named in 64% of 90 answers, the most dominant result we have measured. Real local movers barely surfaced, and a truck-rental company got recommended as a mover.
See the full ranking →Daycare, Astoria: a national chain leads
A national franchise was named in 72% of answers, far ahead of anyone local. The neighborhood's real, licensed independent daycares were nearly absent, even the ones with years in the neighborhood and strong reviews.
Lawyers, NYC: the base models lag reality
Asked cold for a personal injury firm, the models named one that dissolved in 2020 more often than either firm that replaced it. The old name still dominates the web, so the base models repeat it. A browsing assistant would likely catch the update; these did not. That is how a stale footprint outranks a current one.
Source: Espin Local AI Visibility Index, espinlocal.com/ai-visibility. Journalists and researchers: cite freely with a link. Data license CC BY 4.0.
Browse the full ranking by trade and town
Each page carries the verified names and the published method for that market. New towns and trades added as we measure them.
The numbers only matter if you can trust how we got them.
The per-business scoring behind our reports is public too: how I score.
Want to know if AI names you? Find out free.
The free visibility snapshot asks the assistants about your business the same way, now with live web search on, checks your Google profile, reviews, and listings, and shows where you stand and what to fix first. NYC and Long Island, any trade. Wondering what drives it? Start with whether your business shows up in AI search.