Can ChatGPT find NYC small businesses?

East Village: 874 sites, 38% weak. West Village: 793 sites, 42% weak. Chelsea: 699 sites, 13% weak — and a story in why.

Often not, and the gap is measurable street by street. Our scan corpus includes 2,366 Manhattan small-business websites across three neighbourhood cuts — the East Village (874 sites), the West Village (793) and Chelsea (699) — plus 156 in Edison, NJ. In the two Village cuts, roughly 4 in 10 sites score under 50 on the signals AI assistants actually read: 38% in the East Village, 42% in the West Village, 44% in Edison.

The neighbourhood numbers

RegionSitesAvg healthUnder 50Schema markupSPF
East Village, Manhattan8745238%57%63%
West Village, Manhattan7935142%53%68%
Chelsea, Manhattan6995813%24%25%
Edison, NJ1565144%49%53%

Chelsea is the interesting row: the fewest weak sites, but schema-markup adoption of just 24% and a mobile-viewport rate of 36%, against 57% and 89% in the East Village. Different neighbourhoods fail in different ways, which is why each cut gets its own page in the directory.

What being invisible to an assistant looks like

SOCi's study of 350,000 locations found just 1.2% of local businesses are ever recommended by ChatGPT, against 35.9% visible in Google's local pack — while BrightLocal's 2026 research found 45% of consumers now use AI to find local businesses. A site the assistant cannot read is not penalised; it simply is not among the candidates when a visitor asks for a wine bar in the East Village. It is a site the question never reaches.

What a NYC business can check today

Four checks cover the common failures: view your homepage with JavaScript disabled and see whether your name, address and hours survive; read your robots.txt for AI crawlers blocked by name; look for application/ld+json in your page source; and search your business name plus neighbourhood in an assistant. Or run the free scan — no account, results in about a minute — and get all of it measured against the same checks these neighbourhood aggregates were built from.

Straight answers

Which Manhattan neighbourhood scored best?

Chelsea, by average health score — 58 out of 100 with only 13% of its 699 scanned sites under 50. But its schema-markup adoption was the lowest of the three Manhattan cuts at 24%, so its strength is fundamentals, not structured data.

How many NYC business websites did you scan?

The published Manhattan cuts cover 2,366 sites: 874 in the East Village, 793 in the West Village and 699 in Chelsea. A further 156 sites were scanned in Edison, NJ. Aggregates only, no business named.

Is my neighbourhood in the data?

The corpus publishes a cut for any region with at least 30 scanned sites; today that is the three Manhattan neighbourhoods and Edison, NJ. Whatever your neighbourhood, the free scan measures your own site against the same checks.

Does being on Google Maps mean ChatGPT can find me?

No — they are different indexes read in different ways. SOCi's 350,000-location study found 35.9% of local businesses visible in Google's local pack but only 1.2% ever recommended by ChatGPT.

Related

Where does your site stand? The scan is free, takes about 60 seconds, and needs no account. It fetches your homepage exactly the way GPTBot, ClaudeBot and PerplexityBot do and tells you what they receive.