Updated 8 September 2026. An earlier version of this study reported that 10% of the sites did not load. That figure was wrong: our checker counted bot-protection responses as failures. The numbers below are the corrected ones, and the methodology section explains exactly what changed.
We took 420 independent local businesses across Florida — sixteen cities, sixteen trades, no chains — and ran the same automated check against every one of their websites. Not the ones we were pitching. Not the ones someone complained about. A random sample, audited start to finish, with every result kept whether the site passed or failed.
Twelve of them came back clean.
Twelve. The average site had 2.4 things wrong with it, and almost none of those things were about design, budget, or how the site looks to a human being. They were about whether a machine can read it — which is now most of the traffic, because most of the traffic starts as a search or a question put to an assistant.
What we found
Of the 420 sites, 23 were genuinely broken and 18 were closed to automated requests. The percentages below are drawn from the 379 we could actually read.
| Problem | Share |
|---|---|
| No structured markup (LocalBusiness or similar) | 46% |
| No page description — Google writes one for you | 19% |
| No address anywhere on the page | 19% |
| Copyright line still says 2023 or earlier | 16% |
| Page weighs more than 600 KB before images | 13% |
| Title tag too short to say anything | 11% |
| No phone number in the page source | 10% |
| Renders only in the browser — crawlers get a blank page | 7% |
| Still served over plain HTTP | 6% |
| No Spanish version | 90% |
Separately, 5.5% of the websites in the sample were dead — expired certificates, dead hosting, domains that no longer resolve.
That last table row deserves its own sentence. Ninety percent of Florida local business websites exist in one language, in a state where a very large share of customers search in another. Google will not translate you into a result you do not have.
The one that costs the most
Structured markup is the finding worth understanding, because it is invisible and it is free.
Markup is a small block of machine-readable text that states plainly what your business is, where it is, when it is open, and how to reach it. Humans never see it. Google Maps reads it. So do the AI assistants that now answer a growing share of "who should I call about this" questions — and when they answer, they answer from sources they can parse. A site with no markup is not penalised. It is simply not quotable. The assistant reaches for a directory listing or a review site instead, and your competitor's name comes out of the speaker.
Nearly half of Florida local businesses are in that position right now. Adding the markup takes an hour of a developer's time, or is a checkbox in most site builders.
One in eighteen websites is simply gone
Twenty-three of the 420 sites failed to respond at all. These are operating businesses. They have listed phone numbers and recent reviews. Their websites are dead, and nobody has noticed, because the owner has the page bookmarked in a browser that shows a cached version, or has not typed their own address in a year.
Another 7% technically load but render entirely in JavaScript. Open one in a browser and it looks fine. Fetch it the way a crawler does and you get an empty shell. From Google's side and from an assistant's side, those are close to the same thing as being offline.
It is not about how big or how loved you are
We expected the well-established businesses to look better. They do not. Being good at your trade, and being visibly appreciated for it, does not make your website legible to a machine. Those are unrelated skills, and there is no reason a barber or a vet should have the second one.
By trade
Share of each trade with no structured markup:
| Trade | Sites read | No markup |
|---|---|---|
| Dry cleaners | 17 | 82% |
| Auto repair | 28 | 61% |
| Barber shops | 18 | 56% |
| Beauty salons | 20 | 55% |
| Pet grooming | 27 | 52% |
| Insurance agents | 29 | 52% |
| Gyms | 19 | 47% |
| Bakeries | 19 | 47% |
| Day care | 21 | 43% |
| Law firms | 26 | 42% |
| Veterinary clinics | 31 | 39% |
| Plumbers | 26 | 35% |
| Landscaping | 23 | 35% |
| Florists | 31 | 32% |
| Dentists | 32 | 31% |
The pattern is roughly what you would guess: trades that sell through appointments and referrals, and never had a reason to think about search, sit at the top. Trades where somebody once sold the owner a professionally built site — dentists, florists, law firms — sit at the bottom. The difference is not effort. It is whether a competent person happened to touch the site once.
By city
| City | Sites read | No markup |
|---|---|---|
| Orlando | 22 | 55% |
| West Palm Beach | 26 | 54% |
| Coral Gables | 24 | 50% |
| Boca Raton | 28 | 50% |
| Doral | 25 | 48% |
| Tallahassee | 31 | 48% |
| Tampa | 30 | 47% |
| Jacksonville | 24 | 46% |
| Hollywood | 16 | 44% |
| Fort Lauderdale | 27 | 41% |
| Gainesville | 25 | 40% |
| Naples | 41 | 37% |
| Sarasota | 25 | 24% |
City samples rest on 16 to 41 sites each and should be read as signals, not measurements.
What to check on your own site, in order
None of this needs an agency. It needs twenty minutes.
- Open your website in a private browser window, typing the address by hand. Not from a bookmark, not from your history. One site in eighteen failed here.
- Search Google for your business name. Read the grey text under the link. If it is a fragment of your navigation menu or a sentence you do not recognise, you have no page description and Google is writing one for you.
- Check that your address and phone number appear as text on the page — not only inside an image, and not only on a contact page behind a click.
- Ask whoever maintains the site whether it has LocalBusiness structured data. If the answer is confusion, that is your answer. Google's Rich Results Test will tell you in ten seconds for free.
- Look at the copyright line in your own footer. If it says 2023, every visitor has been quietly told the business may not be running.
- If you serve Spanish-speaking customers, check whether your site does. Nine in ten in this state do not.
Fixing all six is a day of work for a competent developer and costs less than a month of the advertising most owners are already buying to compensate for the problem.
How we did this, and what we got wrong along the way
We pulled 4,208 Florida businesses with websites from Google's Places API across sixteen cities and sixteen trades. We removed national chains and franchises — any domain appearing more than three times — along with businesses whose "website" was a Facebook, Instagram, or booking-platform page, leaving 3,453 genuinely independent operators. From those we drew 420 at random with a fixed seed, and audited every single one, keeping the result whether the site passed or failed.
We have corrected this study twice, and both corrections are worth stating plainly.
The first version used the wrong sample. It drew on our own prospecting pipeline, which by design only records a business once its website has already failed a screen. The percentages were real but the sample was pre-selected for breakage, and the headline came out at 67% instead of the low forties. We replaced it with the random sample described above.
The second version used a checker that was too crude. It reported that 10% of sites did not load. When we opened those sites by hand, most of them worked: they were returning a "forbidden" response to our automated request, which is ordinary bot protection, not a broken website. We rebuilt the checker to retry before giving a verdict and to record bot protection as its own category. Of 420 sites, 18 are closed to automation and 23 are genuinely dead — 5.5%, not 10%.
A biased sample and a crude instrument both err in favour of whoever is holding them, which is exactly why they feel so quotable.
What this study still does not tell you: it is a snapshot of sixteen cities, weighted toward businesses Google surfaces for a category search, so very small operators with weak listings are underrepresented. The checks are automated and conservative — a site could carry structured data in a form our parser missed, which would make the true markup figure slightly better than 46%, not worse.
The dataset and the script are available on request. We would rather be checked than believed.
The short version
Twelve of 420 Florida local business websites were free of basic, mechanical faults. Nearly half cannot be quoted by a map or an assistant. One in eighteen does not open. And almost none of it is a design problem, a budget problem, or a sign that the owner is bad at their job — it is a category of maintenance nobody told them existed.
If you want to know which of these apply to you specifically, we do a free five-page audit and send it within 48 hours. No call unless you ask for one.
