Local Business AI Readiness Study: 288 Sites

Short answer: Most local business websites are only partly readable by AI assistants. In September 2026 we ran Alphaa's 23-point AI-readiness check on 288 real local business websites: 17% turned away an automated reader entirely, 71% of the readable sites had missing or incomplete structured business facts, 46% didn't show a phone number, address or hours in text AI can read, and fewer than 2% passed 20 of the 23 checks.

What did we check, and on which businesses?

We checked 288 websites belonging to dentists, plumbers and HVAC companies, law firms, veterinarians and restaurants in Austin, Denver, Phoenix, Miami, Seattle and Chicago. The businesses came from OpenStreetMap (public listings that include a website), sampled at random — up to 60 per industry — and each site was checked once on September 25, 2026.

• The check: the same 23 checks Alphaa runs in its free 60-second AI check — whether AI crawlers can get in, whether the content is readable without JavaScript, structured data, phone/address/hours, FAQ content, reviews readable as text, profile links, freshness and more.
• How: plain HTTP requests for the homepage, robots.txt, sitemap, llms.txt and one inner page, plus homepage requests using the ChatGPT, Claude and Perplexity crawler identities.
• Limits: one snapshot, one network location, homepages only for most checks. A firewall can treat our requests differently from the real crawlers, so "blocked" means likely blocked. Bing verification by DNS can't be seen from outside, so that figure is an upper bound.

How many local business websites can AI actually read?

About four in five: 239 of the 288 sites (83%) returned a readable homepage, while 49 (17%) refused an automated visitor with an error or a bot challenge. A site that turns away automated readers is invisible to any AI assistant that gets the same treatment, so the assistant falls back on what other websites say about the business.

• Plumbing and HVAC sites turned readers away most often: 15 of 48 (31%).
• Of the 239 readable sites, 17% served a block or challenge to at least one AI crawler identity (ChatGPT, Claude or Perplexity) while letting a normal browser in.
• 10% showed almost no readable text before JavaScript runs — and most AI crawlers don't run it.

What are local websites most often missing?

The most common gaps are the ones AI relies on to describe and trust a business: complete structured facts, FAQ answers, reviews in plain text and fresh content. Among the 239 readable sites:

Two failures deserve a note. Google says its AI features don't require an llms.txt file; other assistants can use it, which is why we check it, but it's the least critical item on this list. And 88% showed no sign of Bing Webmaster Tools — worth fixing because ChatGPT search results have been found to overlap heavily with Bing's — but some of those sites may be verified by DNS, which we can't see.

Which industries are most ready for AI search?

Law firms, dentists and plumbers scored about the same — roughly 15 of 23 checks — while restaurants scored lowest at 11 of 23. Restaurants were the most likely to lack structured data, readable reviews, FAQ answers and profile links, which matters because diners ask AI for recommendations constantly.

How many sites are fully ready?

Almost none: only 4 of the 239 readable sites (under 2%) passed 20 or more of the 23 checks, and the median site passed 15. In practice that means most local businesses are leaving AI to guess — or to recommend the competitor whose facts are easier to read.

What should a local business fix first?

Fix access first, then facts, then trust: make sure AI can read your site at all, put your phone, address and hours on the homepage in plain text and in structured data, then add FAQ answers and a few real reviews as text.

1. Check that your host or firewall isn't turning away AI crawlers, and that your homepage text loads without JavaScript.
2. Put your name, phone, address, hours and service area on the homepage — as text, not an image.
3. Add LocalBusiness structured data with address, phone, hours, map location and links to your Google, Yelp and social profiles.
4. Publish answers to the five questions customers ask before they call.
5. Show a handful of real reviews as text, not only inside a widget.

The key takeaway is that the gap is rarely exotic: most local sites fail on basics — access, facts and plain-text trust signals — that AI needs before it will recommend anyone, and those are fixable in a week.

What else do people ask about the AI readiness study?

How were the 288 sites selected?

They were drawn across common local service categories and checked with the same automated tests, so the comparison between industries is like for like rather than cherry-picked.

Does a low readiness score mean a business is invisible?

No. Readiness measures what an engine can read from the site itself, and many businesses are still named from their profile and reviews alone.

Which single failure was most common?

Key facts not present in crawlable text, usually because pricing, service areas or hours lived only in an image, a PDF or a client-side widget.

Will you repeat the study?

That is the intention, with the same checks so the numbers stay comparable. We will state any change in method rather than quietly adjusting it.

How do you see where your own site stands?

Run the same 23 checks on your site, plus a live question to ChatGPT, Gemini, Claude and Perplexity, with the free 60-second check. For what each check means, see the AEO checklist and why JavaScript-only pages are invisible to AI.