Raqzor research · September 2026
Can AI read Florida’s local businesses?
We took the top Google Maps results for 10 trades in 40 Florida cities, from Pensacola to Key West, 7,834 listings in all, and ran each business’s website through the same check as our free tool. Here’s what ChatGPT, Google, Perplexity and other AI assistants can learn from those websites, and what’s missing.
- 60%have no questions and answers on their homepage
- 46%don’t label their name, address and phone for machines
- 1 in 12top-ranked businesses has no website at all
- 8%pass every check in our free tool
Shares of the 5,827 businesses whose homepage we could check, except the website figure, which counts all 7,596.
85%
of the 6,873 businesses with a website showed our checker their homepage
First, can the website be opened at all?
Before an AI assistant can learn anything from a website, a crawler has to open it. For the other 15%, our checker didn’t get as far as the homepage.
Most of those answered with a security challenge instead of the page, usually from the web host’s protection against automated visitors. The host behind most of them, SiteGround, says it lets the crawlers behind ChatGPT, Claude and Perplexity through by default and blocks only crawlers that collect AI training data. For the rest, whether a crawler gets through depends on settings nobody can see from outside, so we don’t count any of these websites as blocked. If yours is one of them, ask your host which crawlers its protection lets in.
The remaining 3.9%: 128 forwarded to a different website, 61 returned an error page, and 74 had a robots.txt that asked crawlers to stay out or failed to load. A robots.txt that errors tells well-behaved crawlers to skip the whole site.
48
of 6,873 businesses with a website block an AI assistant’s crawler in robots.txt
Almost nobody blocks AI in robots.txt
A common worry is that a website shuts out ChatGPT. In robots.txt, the file that tells crawlers where they may go, that’s rare: 48 of the 6,873 businesses with a website block at least one AI assistant’s search crawler. 33 of them block ChatGPT’s, and 23 of those shut out every crawler, Google’s included. Among the businesses we could check, 22 more limit an assistant without blocking it, and 1.5% opt out of AI training, which doesn’t take them out of AI answers.
Accidents are more common. 137 of the businesses we could check (2.4%) have a homepage that’s nearly empty to a crawler that doesn’t run JavaScript, and several AI crawlers don’t. 55 carry a noindex tag that tells search engines, and the AI answers built on them, to leave the page out. In all, 3.4% of the businesses we could check have a problem on the homepage that can keep AI out entirely.
46%
don’t label their business details for machines
Business details hidden from machines
Structured data is a set of hidden labels that states, in a form machines read reliably, who a business is, where it is and how to reach it. 46% of the businesses we could check don’t give their name with an address or phone number this way on their homepage, and 20% have no structured data there at all.
It isn’t required, and plenty of businesses rank without it. But it was more common at the top: 60% of the top three Google Maps results labeled their details, against 53% further down. That doesn’t prove the labels moved anyone up. It does show what the leaders tend to have.
60%
have no questions and answers on their homepage
Plenty of words, few answers
People ask AI assistants questions, and assistants tend to quote pages that answer the exact question asked: what a repair costs, whether you take emergency calls, which insurance you accept. 60% of the businesses we could check have no questions and answers on their homepage at all: no FAQ and no question-style headings. Another 22% answer only one or two. Our free check looks for at least three.
It isn’t for lack of text. The median homepage has about 913 words. They just aren’t shaped as answers.
47%
showed a rating on Bing at least a star lower than on Google
Bing tells a different story
Microsoft Copilot answers from Bing. We ran 40 of the same searches on Bing, 10 trades in Sarasota, Tampa, Fort Myers and Naples. Its local results listed a median of 13 businesses per search, and only 50% of them were in Google’s top 20 for the same search.
The ratings differ too. Bing shows ratings from other sites, most often Yelp (65% of its listings), and 29% showed none. For the 211 listings that appeared on both with a rating on each, the median was 4.8 stars on Google and 4.0 on Bing, and 47% were at least a full star lower on Bing. If you only watch your Google reviews, you may not know what Copilot sees.
95%
of recent reviews got an owner reply, for the median #1 listing
What the top of the map does with reviews
Reviews aren’t on your website, but they’re among the first things an assistant finds when it looks a business up. We looked at the newest reviews, up to 60 each, of 40 leaders: the #1 Google Maps result for each of the 10 trades in Tampa, Sarasota, Miami and Orlando.
The median leader replied to 95% of them, and 24 of the 40 replied to at least nine in ten. The median leader gets about 9 new reviews a month. It’s a small sample, so read it as a picture of the leaders, not a rule.
43%
of the businesses we could check have their website on Cloudflare
Two more things we noticed
Cloudflare’s September change. Cloudflare announced that from September 15, 2026, sites that block AI training in its settings also block Google’s, Bing’s and Apple’s crawlers. That setting isn’t visible from outside. None of the robots.txt files we read carried Cloudflare’s AI-training opt-out, which is a good sign, but it can’t rule the setting out. If your site is on Cloudflare, open AI Crawl Control in your dashboard and make sure search crawlers are allowed.
llms.txt is spreading on its own. 40% of the businesses we could check have an llms.txt file, a plain summary written for AI tools. Several of the ones we opened said they were generated automatically by an SEO plugin such as Yoast or Rank Math. It’s optional and few AI crawlers read it yet, so it isn’t worth worrying about either way.
By trade
Injury lawyers and roofers did best: they pass every check most often, and more of them answer questions on the homepage than most trades. Med spas answer the fewest. Auto repair stands out for another reason: 22% of the top-ranked shops have no website of their own, against 1.6% of injury law firms.
| Trade | No website | No business labels | No questions answered | Every check passed |
|---|---|---|---|---|
| Injury lawyers | 2% | 40% | 43% | 15% |
| Roofers | 7% | 43% | 55% | 13% |
| Plumbers | 11% | 47% | 59% | 8% |
| Auto repair | 22% | 45% | 63% | 8% |
| Pest control | 7% | 52% | 47% | 8% |
| AC repair | 7% | 43% | 59% | 7% |
| Dentists | 3% | 40% | 68% | 6% |
| Electricians | 14% | 53% | 62% | 5% |
| Med spas | 7% | 51% | 72% | 5% |
| Pool service | 14% | 50% | 68% | 4% |
By region
North Florida and the Panhandle stand out. 12% of its top listings have no website, more than twice the share around Tampa Bay, and more than half of those we could check don’t label their business details. Central and North Florida also answer fewer questions than Tampa Bay and South Florida. Gaps of a point or two are within the noise of a sample this size.
| Region | No website | No business labels | No questions answered | Every check passed |
|---|---|---|---|---|
| Tampa Bay & Southwest Florida | 6% | 43% | 57% | 8% |
| South Florida | 9% | 44% | 56% | 10% |
| Central Florida | 10% | 45% | 62% | 7% |
| North Florida & Panhandle | 12% | 53% | 63% | 7% |
What to fix first
Make sure crawlers can open your homepage
If your host or security service challenges automated visitors, ask which crawlers it lets in and make sure search and AI crawlers are among them. On Cloudflare, that’s AI Crawl Control.
Answer the questions customers ask
Add a short FAQ to your homepage: prices or how you quote, service area, hours, emergencies, insurance. Two or three plain sentences per answer.
Label your business for machines
Add schema.org business details, using the closest type (LocalBusiness, Plumber, Dentist and so on) with your name, address, phone and hours. Most website builders and SEO plugins can do this for you.
Look yourself up on Bing and Yelp
Claim your Bing Places listing and read what Yelp shows about you. Copilot answers from Bing, and Bing often shows Yelp’s rating, not Google’s.
Reply to your reviews
In our sample, the median #1 Google Maps listing replied to 95% of its recent reviews.
How does your website score?
Run the same check we used here. It takes about 20 seconds, it’s free, and it tells you what to fix in plain English.
Questions about this report
How many Florida local business websites can AI read?
In our sample of 7,596 top-ranked businesses in 40 Florida cities, our checker could read the homepage of 85% of the businesses with a website. Of those, 3.4% had a problem that can keep AI assistants out entirely, and 8% passed every check in our free tool.
Do local businesses block ChatGPT or other AI assistants?
Rarely. Only 48 of the 6,873 businesses with a website blocked an AI assistant’s search crawler in robots.txt. Accidental problems were more common: 137 homepages were nearly empty without JavaScript and 55 asked search engines not to index them.
What is the most common gap?
Answering questions. 60% of the businesses we could check had no questions and answers on their homepage at all. Next came structured business details: 46% didn’t label their name with an address or phone number for machines.
Is this report about Google Maps rankings?
No. We used Google Maps to choose which businesses to look at, then checked their websites. Which businesses an AI assistant recommends depends on much more than a website, and this report doesn’t measure what assistants say.
How we did it
What we collected
We collected the top Google Maps results, up to 20 per search, for 10 searches: plumber, air conditioning repair, roofing contractor, electrician, dentist, med spa, personal injury lawyer, auto repair shop, pest control and pool service. We ran each in 40 cities, 10 in each of four regions: Tampa Bay & Southwest Florida (Bradenton, Clearwater, Fort Myers, Lakeland, Naples, Port Charlotte, Sarasota, St. Petersburg, Tampa, Venice); South Florida (Boca Raton, Coral Springs, Fort Lauderdale, Hollywood, Homestead, Key West, Miami, Pompano Beach, Port St. Lucie, West Palm Beach); Central Florida (Daytona Beach, Deltona, Kissimmee, Leesburg, Melbourne, Ocala, Orlando, Sanford, The Villages, Winter Haven); North Florida & Panhandle (Crestview, Destin, Fort Walton Beach, Gainesville, Jacksonville, Lake City, Panama City, Pensacola, St. Augustine, Tallahassee). We collected the Tampa Bay and Southwest Florida cities on September 24, 2026, and the rest on September 25. 17 of the 400 searches, mostly for med spas, returned fewer than 20 businesses. That’s 7,834 listings and 7,596 different businesses, 6,873 of them with a website. For Sarasota, Tampa, Fort Myers and Naples we also collected Bing’s local results for the same 10 searches (506 listings; Bing returned between 3 and 26 per search).
How we checked the websites
Each website’s homepage went through the same code as our free check at raqzor.com/check, once, from one computer on an ordinary internet connection, identifying itself as RaqzorCheck. Before opening a homepage we read the site’s robots.txt, and we skipped any site that asked crawlers to stay out or whose robots.txt failed to load. A website listed for several businesses or searches was checked once.
What the numbers mean
- Figures about websites are shares of the 5,827 businesses whose homepage we could read, “the businesses we could check”, unless we say otherwise. Each business counts once, so a website shared by several listed businesses counts once for each.
- Chains are included. Setting aside the 165 websites that appeared in three or more of the 40 cities moves the main figures by 1.3 points or less.
- The check reads homepages only. It shows what AI can read, not whether any assistant names a business.
- Security settings can treat our checker differently from Google’s crawlers or an AI company’s, so a website that turned us away may still let them in.
- The review figures come from 40 businesses and describe those businesses only.
What we didn’t do
We didn’t contact any business, this report doesn’t name any, and we don’t publish the list. Questions about the method: [email protected].