Between March and August 2026 we ran automated technical audits on local business websites across Fairfield County, Connecticut and Westchester County, New York. This is what we found, with the methodology stated in full so you can judge the numbers rather than take them on trust.
of audited local business websites have no LocalBusiness structured data — the markup that tells Google, ChatGPT, Perplexity and Google's AI Overviews what the business is and where it operates. 484 of 622 sites.
These are binary presence-or-absence checks, not judgement calls, and all come from a single audit pipeline — so they are directly comparable to one another.
| What we checked | Missing | Sites |
|---|---|---|
| No LocalBusiness structured data The markup that tells Google — and ChatGPT, Perplexity and AI Overviews — what kind of business this is, where it operates, and how to contact it. Without it, an assistant answering "best dentist near me" has to infer everything from prose. | 77.8% | 484/622 |
| No click-to-call link On a phone the visitor has to copy the number by hand. For a service business where the call is the conversion, that is a self-inflicted wound. | 51.0% | 317/622 |
| Missing meta description Google writes the search snippet for them, from whatever text it finds first. | 30.4% | 189/622 |
| No sitemap.xml Crawlers discover pages by following links and luck rather than being handed a map. | 27.0% | 168/622 |
| No robots.txt Usually harmless, occasionally catastrophic — it is also where AI crawler access is granted or denied. | 12.9% | 80/622 |
| No mobile viewport meta tag The <meta name="viewport"> tag that tells a browser to render the page at device width. Without it a phone renders the desktop layout scaled down. This is the specific, checkable thing we measured — not a general judgement about whether the site looks good on a phone. Rarer than the industry implies: the mobile problem is largely solved. | 7.7% | 48/622 |
| Missing <title> The single most basic on-page tag, absent on roughly one site in twenty. | 4.5% | 28/622 |
| No HTTPS Nearly extinct, and the remaining cases show a browser security warning. | 2.4% | 15/622 |
The pattern is consistent and, we think, counterintuitive. The basics are largely solved — only 7.7% of sites are not mobile-optimised and just 2.4% still lack HTTPS. Local businesses fixed the 2015 problems. What almost none of them have done is the work that decides whether software can understand the business at all.
A human reading a dental practice's homepage knows instantly that it is a dental practice in Norwalk that opens at eight. A search engine or an AI assistant has to infer all of that from prose unless the page states it in machine-readable form.
That inference was tolerable when search returned ten blue links and the customer did the rest. It is much less tolerable now that a growing share of "near me" questions are answered by an assistant that names two or three businesses and stops. There is no page two in an AI answer. Being legible to the software is increasingly the whole game, and more than three quarters of local businesses here are not.
The flip side is that this is cheap to fix and almost nobody has. A correct LocalBusiness block is perhaps an hour of work.
Mean technical score out of 100, for towns where we audited at least 20 sites. Towns with smaller samples are omitted rather than reported with false precision.
| Town | Mean score | Sites audited |
|---|---|---|
| New Canaan, CT | 83.2 | 22 |
| Greenwich, CT | 82.5 | 55 |
| Westport, CT | 81.9 | 21 |
| Darien, CT | 80.2 | 87 |
| Norwalk, CT | 75.9 | 50 |
| Fairfield, CT | 75.9 | 28 |
| Stamford, CT | 75.6 | 52 |
| Bridgeport, CT | 73.1 | 24 |
The honest reading of this table is that the differences between towns are small — about ten points from New Canaan at the top to Bridgeport at the bottom, which is less than the spread we see within any single town. Wealthier towns score slightly higher, but not dramatically, and no town is doing well on structured data. If you are looking for a story about affluent towns dominating local search, this data does not support it.
Why Darien has so many sites in this table. Darien's sample of 87 is larger than Greenwich's or Stamford's, which is an artefact of our crawl rather than a fact about Darien. Darien actually has the smallest prospect pool of the large towns here — 197 collected domains, against 323 for Greenwich, 320 for Norwalk and 396 for Stamford. It has the largest audited sample because a much higher share of that pool was crawled: about 45% of Darien's domains were audited, against 9–17% for every other town in the table, most of it in a single batch on 11 August 2026. The practical effect is that Darien's mean is estimated more precisely than its neighbours', not that Darien has more businesses or was selected for any property of its websites.
Sample: 622 local business websites in Fairfield County CT and Westchester County NY, audited June–August 2026.
Collection: one automated pipeline requesting each site over four host and scheme variants with a retry, then parsing the served HTML. Sites we could not reach are excluded, never scored zero — an unreachable site is an unknown, not a bad one. Of the 703 domains we attempted to crawl, 75 could not be reached — 50 DNS or connection failures, 2 timeouts and 23 other HTTP errors — and were excluded on that basis. Removing duplicate domains, where several businesses shared a single website, leaves the 622 sites analysed here.
Scoring. Each site starts at 100 and loses points for specific, checkable failures:
<title> — −10sitemap.xml — −5robots.txt — −5The deductions total 100, so a site failing every check scores 0. The town score is the unweighted mean of all scored sites in that town. It measures technical hygiene, not marketing quality, and it is not a judgement about the business.
Town coverage. Town figures are reported only where n ≥ 20, which accounts for 339 of the 622 sites. The remaining 283 are spread across 43 further towns that each fell below that threshold — roughly 170 in Fairfield County and the rest in Westchester. The largest of them are Port Chester NY and Milford CT (16 sites each), Danbury CT and Monroe CT (15), Trumbull CT (14), and White Plains NY, Stratford CT and Ridgefield CT (13). They are omitted from the table rather than reported with false precision, but they are included in every county-wide figure above.
Known limitations. We do not execute JavaScript, so a site whose content loads client-side is recorded as unreadable and excluded from the text-based checks rather than marked deficient. The sample skews toward established businesses, because prospects are sourced by business email domain — sole traders using a personal address are underrepresented. That makes the structured-data figure conservative: even among businesses running their own mail domain, more than three in four are missing it.
Survivorship bias. Every site in this sample is a site that exists and answered. Businesses with no website at all — a meaningful share of the smallest local firms — cannot appear here, and they are precisely the businesses least likely to have structured data or a sitemap. The gaps we report are therefore a floor, not a ceiling. Nor is the sample randomly drawn: it is the set of domains our prospecting pipeline had already collected, so town-level sample sizes reflect our crawl scheduling rather than the true number of businesses in each town.
Conflict of interest. Stamford AI Consulting sells the work this report implies is needed — structured data, technical SEO and AI-visibility fixes for local businesses. We chose the checks, the weights and the thresholds, and we benefit commercially if you conclude your site is deficient. Every underlying check is a mechanical, independently verifiable property of a page: you can confirm any figure here yourself by viewing source, or have someone with no stake in the answer do it. We would rather you verify than take our word for it.
Published August 2026, v1.1 — methodology section updated 17 August 2026 to add the exclusion count, the full scoring formula, town coverage below the reporting threshold, and the limitations above.
In rough order of return on effort, based on what the data actually shows:
622 local business websites across Fairfield County, Connecticut and Westchester County, New York, where our crawler successfully retrieved the site. We attempted 703 domains in total; 75 could not be reached — 50 DNS or connection failures, 2 timeouts and 23 other HTTP errors — and were excluded from every figure rather than counted as failures. Duplicate domains, where several businesses shared one website, were also removed.
77.8% — 484 of 622 audited sites had no LocalBusiness structured data. This is the single most common gap we found, and the one that matters most for whether AI assistants can identify and recommend a business.
It is a block of machine-readable markup that states what the business is, where it operates, its hours and its contact details. Search engines and AI assistants read it directly instead of guessing from page copy. Without it a business is legible to humans but ambiguous to software.
New Canaan, at a mean technical score of 83.2 across 22 audited sites, followed by Greenwich at 82.5 and Westport at 81.9. Bridgeport was lowest at 73.1. The spread between the best and worst town is about ten points — smaller than the spread within any single town.
Each site starts at 100 and loses points for specific, checkable failures: no HTTPS −20, no LocalBusiness structured data −15, no mobile viewport meta tag −15, slow first byte over 2.5 seconds −15, missing meta description −10, missing <title> −10, no click-to-call link −5, no sitemap.xml −5, no robots.txt −5. The deductions total 100, so a site failing every check scores 0. A town's score is the unweighted mean of all scored sites in that town. It is a technical hygiene score, not a judgement about the business or its marketing.
Yes, free. We run the same checks plus a review of your Google Business Profile, reviews and closest competitors, and send back a scored report. No sales call.
Same technical checks as this study, plus your Google Business Profile, reviews and closest competitors. A scored report, no sales call.
Get the free audit →Questions about the data? Call (203) 651-8786 or email info@stamfordaiconsulting.com.