Original data · August 2026

We Ran 17 Local Businesses Through an AI-Visibility Check — Most Aren't Ready to Be Recommended by AI. Here's the Data.

By the AI Did It For You team · Published August 24, 2026 · ~7 min read

We took 17 real US local businesses, ran each one through an automated AI-visibility scanner, and recorded what came back. Only 5 of 17 (29%) were rated “AI-ready.” 12 of 17 (71%) were partially or largely invisible to AI search, and 7 of 17 (41%) landed in the worst bucket — “largely invisible.” The average score was 63 out of 100. Below is the exact method (so you can replicate it) and the full data table.

The question we wanted to answer

When someone asks ChatGPT, Google's AI Overviews, Perplexity or Claude for “the best dentist in Boise” or “a good HVAC company near me,” a handful of businesses get named and the rest don't exist as far as the answer is concerned. Being named is quickly becoming as important as ranking on Google used to be.

So we asked a simple, testable question: when AI has to identify and recommend a local business, how ready is the average small business to be picked? Or, put the other way: how invisible are ordinary local businesses to AI search right now?

The method (so you can replicate it)

This is a deliberately simple, repeatable procedure. Anyone can run the same steps and check our numbers.

• Sample: 3 categories × ~6 businesses each — general/family dentists in Boise, ID, personal-injury & family-law lawyers in Tucson, AZ, and HVAC contractors in Omaha, NE. Businesses were pulled from the first page of ordinary Google searches (a convenience sample).
• Instrument: the free AI Visibility Scanner. For each business we entered its real homepage URL, its trade, and its city.
• What we recorded: the scanner's 0–100 AI-visibility score and its verdict — RED (largely invisible), YELLOW (partially visible), or GREEN (AI-ready).
• Date: all scans were run on 24 August 2026, one snapshot in time.
• Failures: if a scan errored (site unreachable / bot-blocked), we noted it and moved on rather than guessing a number.

What the scanner checks: whether an AI model can read the site (crawler access), whether the page carries the structured-data and content signals AI relies on (LocalBusiness schema, a clear title/meta, visible name-phone-location, FAQ content, an llms.txt file), and how clearly the homepage identifies the business, its services and its city. It is an automated estimate of AI-readiness — not a live query to ChatGPT — which matters for how you read the results (see limitations).

The data

Seventeen businesses returned a score. Three more (one per category) blocked the scanner's crawler entirely and could not be scored — itself a form of AI-invisibility, discussed below. Here is every real result:

Aggregate (N = 17): GREEN 5 · YELLOW 5 · RED 7. Average score 63/100. By category, the averages were HVAC 69.8 (n=6), dentists 64.3 (n=6) and lawyers 53.2 (n=5) — the law firms were the least AI-visible group in our sample.

What the numbers mean

Three things stood out, and we want to be careful to separate what the data shows from what we're inferring.

1. Most local sites aren't built for AI to read them. The recurring gaps were the same across all three trades: no LocalBusiness structured data, no FAQ content, and — almost universally — no llms.txt file. These are exactly the machine-readable signals AI systems use to identify a business and decide whether to name it.

2. Being “identifiable” isn't the same as being AI-ready. Several sites the scanner could clearly describe still scored RED. Boise River Dental, for example, was easily identified as “a full-service dental practice... led by Dr. Travis Weimer” and still scored 45, because the structured-data and llms.txt signals were missing. Readability by a human doesn't guarantee an AI will confidently recommend you.

3. The sample is probably better than average — and it's still this bad. Every business here was pulled from Google's first page. If anything, these are the more visible local businesses, not the least. That the top-of-Google set still comes out 71% not-AI-ready suggests the true population is likely worse.

The most striking individual cases

• The perfect score: Lamb Family Dental hit 100/100. The scanner noted it could “immediately identify” the practice, its named dentists, address, services and reviews — “highly recommendable” for a “Dentist in Boise, ID” query. Proof this is achievable, not theoretical.
• The block-page problem: Blaisdell Family Dentistry scored 25 because its homepage returned a Cloudflare block page — “zero dental or Boise-specific content, making it impossible for an LLM to identify the business.” Aggressive bot-blocking can make you invisible to AI without you realizing it.
• Empty to a machine: two Tucson law firms scored 25 with the same verdict — the homepage “contains no usable content about legal services, location, or attorney credentials.” A visually rich, JavaScript-heavy site can still read as blank to an AI crawler.
• The three that blocked us: one business in each category (a dentist, a law firm and an HVAC company) refused the scanner's crawler outright. If an automated agent can't reach your homepage, neither can most AI systems.

How to check your own business (in about 30 seconds)

You don't have to take our word for it — run your own homepage through the same instrument we used:

• Open the free AI Visibility Scanner, enter your URL, trade and city, and read the 0–100 score plus the fix-it checklist.
• Want a broader brand-level look? The AI Visibility Audit checks how your brand shows up more generally.
• Curious how AI citation actually works? See our explainer on getting cited by AI assistants (AEO) and our roundup of the best AI-visibility tools in 2026. The full free AI toolkit is here.

Honest limitations

We'd rather under-claim than oversell, so read the numbers with these caveats:

• One tool, one moment. Everything here is a single scanner's estimate on 24 August 2026. Scores can move as sites (and AI models) change.
• Small, convenience sample. N = 17 across three categories and three cities, chosen from Google's first page. This is a directional spot-check, not a representative census.
• An estimate, not a live AI query. The scanner rates whether your site can be identified and recommended by AI; it does not measure whether a specific ChatGPT answer named you on a given day.
• Not a controlled trial. No control group, no randomized selection, no repeated measures. Treat this as an illustrative data point that invites replication — and please do replicate it.

If you run the same procedure on a different set of businesses, we'd love to see your numbers.