THE SHORT ANSWER
Only about 8 in 100 Florida pool service websites let a customer book, log in or pay online: 15 of the 186 sites we crawled on Oct 30, 2026 (8.1%, 95% CI 5–13%). Among the 75 working sites of businesses sampled from Florida’s state registers, it was 3 (4%). An estimated 47% of registered pool businesses have a website (95% CI 34–69%), based on 395 businesses sampled from 16,439.
Key findings
15 of 186
Across all 186 sites we crawled, 15 (8.1%, 95% CI 5–13%) let a customer book, log in or pay online; among the 75 registry sites it is 3 (4%). Every one was checked by hand, and these are floors.
47%
Nearly half of Florida’s registered pool businesses have a website: an estimated 47% (95% CI 34–69%).
32% vs. 18%
Licensed (DBPR) contractors were 1.7 times as likely to have a findable site as businesses with no state license (32% vs 18%).
100%
The basics are solved: all 75 working registry sites loaded over https with a valid certificate, and none had an expired one.
35%
Only 35% have any analytics tag installed, and only 3 of 75 sites have a blog post dated in the last twelve months.
Method
We wanted a number nobody publishes: how far Florida’s pool businesses have come online across the whole registered industry, not only the ones that already advertise. The study has two layers.
Layer 1, the businesses. We built a list of every Florida entity that is either an active DBPR pool/spa contractor license holder or an active Sunbiz corporation, LLC or fictitious name with a pool word in its name (16,439 entities after removing duplicates). From it we drew a stratified random sample of 400 (seed 20261030) over five regions, and weighted each business by its region’s sampling fraction. Five drop out (three license holders without a trade name, two list errors), leaving 395. For each, we looked for a website without using Google or map data: up to 45 domain guesses built from the business name, a DNS lookup, a fetch of the home page, and a check that the page names the business and is in Florida. Every match and doubtful case was checked by hand.
Correcting for the sites we can’t find. A name-to-domain search misses sites whose address doesn’t look like the business name. We measured how much it misses by running the exact same automated search, blind, on pool businesses whose website we already knew. For the 55 of them we could match to a registered legal name in Sunbiz (the closest match to the registry search), it found 41.8% automatically, and 45.5% when doubtful candidates that were the right site count as found; on the trade names of all 167 it found 43.1%. Each observed rate is divided by the recall counted the same way (Section 04); the interval comes from 5,000 bootstrap draws of both the sample and the recall.
Layer 2, the websites. Every site found was crawled with a phone-sized browser (390 × 844) on Oct 30, 2026: the home page plus up to six inner pages and a blog index, respecting robots.txt, at most one request per second per domain. A rule-based extractor records the features in the table below. For comparison we ran the same crawler on 167 pool businesses with their own website from a map-listing export of Oct 27, 2026 (Tampa Bay, Orlando, Fort Lauderdale and Fort Myers).
| Rung | Name | Counts when… |
|---|---|---|
| 1 | Secure | Final page loads over https and the browser accepts the certificate |
| 2 | Own-domain email | At least one email link (mailto) on the business’s own domain; we classify the domain type and never store the address |
| 3 | Machine-readable | Any schema.org markup (JSON-LD or microdata); LocalBusiness reported separately |
| 4 | Connected | Links at least one social profile (Facebook, Instagram, YouTube, LinkedIn, X, TikTok, Nextdoor, Yelp) |
| 5 | Kept current | Footer copyright year is 2025 or 2026 (missing or impossible years count as not current) |
| 6 | Self-service | Online booking (a recognized scheduler), a customer login or portal, or a way to pay online; every positive checked by hand |
The score is a plain count of the six rungs, 0 to 6. The order of the rungs follows the effort involved for an owner, not a claim about which matters most for sales. Site-level percentages are unweighted shares of working sites with Wilson 95% intervals; weighting them by region would move no figure by more than 1.2 percentage points. Differences between the two lists use Newcombe’s score interval; the score difference uses a bootstrap.
Sample and coverage
| n | Note | |
|---|---|---|
| Registry list (DBPR pool/spa licensees + pool-named Sunbiz entities) | 16,439 | 4,104 hold a DBPR license |
| Sampled businesses in the estimate | 395 | stratified by five regions, 60–100 each |
| … with an own website found | 86 | plus 3 on a franchise’s site (not counted) |
| … of which crawled successfully | 75 | 11 failed, 9 of them blocked our crawler |
| Map-listing sites crawled for comparison | 167 | 114 worked; 23 blocked, 12 excluded by robots.txt, 18 offline or errors |
| Registry sites also in the map list | 4 | 4.7% of 86 |
The 75 working registry sites are 72 crawled as part of the registry list plus 3 that were already on the map list and were crawled there with the same crawler on the same day.
Who is in the list, and who isn’t. Routine pool cleaning needs no state license in Florida: the statute exempts cleaning that doesn’t affect the pool’s structure and water treatment without equipment work (§ 489.105, Florida Statutes). So the DBPR license file only covers contractors (classes CPC-A commercial, CPC-B residential, CPC-C servicing, and the older RP registrations). Cleaning-only businesses appear only through Sunbiz, and only if their registered or fictitious name contains a pool word. A business called “Mike’s Services LLC” that cleans pools is invisible to this study; so is a sole proprietor who never filed a name. The name rule alone would catch only 68% of active DBPR license holders with a business name (2,817 of 4,154), which is why we added the whole license file. A hand check of 100 random list entries found 95 clear pool businesses, 3 unclear and 2 errors.
Registered is not the same as trading. A dormant LLC stays “active” in Sunbiz until its next annual report lapses. Dormant entities rightly have no website, so the website rate describes registered entities, not only businesses that are trading; we can’t tell the two apart.
Data vintages. The DBPR file is a public archive.org copy of the state’s download from Aug 10, 2026: the state site sits behind a bot check that refused our scripted download, and we did not try to get around it. Sunbiz data are from Oct 3 and Oct 7, 2026.
Sites we couldn’t read. 10% of registry-found sites and 14% of map-list sites blocked our clearly labeled crawler (“bot walls”); we did not work around that, so they are excluded from site-level figures. If blocked sites are run by more sophisticated vendors, the site-level figures slightly understate maturity. Another 12 map-list sites were left out because their robots.txt disallowed us or could not be read (which the robots.txt standard treats as “keep out”).
Nearly half of registered pool businesses have a website
The honest headline number has two parts. What we could see: 21.6% of registered pool businesses (17.6–25.6%) have a website at an address that looks like their name. What is probably there: an estimated 47%, once you correct for the sites a name search can’t find.
Dot = estimate, bar = 95% interval. The interval covers sampling and recall uncertainty, not the assumptions discussed below.
Source: Registry sample, n = 395, weighted by region; recall test on known sites. Ruberio, Oct 2026.
Counting like for like. The observed rate and the recall must be counted the same way. Our published 21.6% includes 17 doubtful candidates a person approved; without them (only what the search verified by itself) it is 17.5%. The recall test was fully automatic, but it also lists doubtful candidates that were in fact the right site. So the hand-cleaned rate goes with the recall that counts those, and the automatic rate with the automatic recall:
| Counting · recall set | Observed | Recall | Website share |
|---|---|---|---|
| Automatic-only · legal name | 17.5% | 41.8% | 42.0% |
| Hand-cleaned · legal name | 21.6% | 45.5% | 47.4% |
| Automatic-only · trade name | 17.5% | 43.1% | 40.7% |
| Hand-cleaned · trade name | 21.6% | 46.7% | 46.2% |
Our main estimate is the first hand-cleaned row (47.4%, 95% CI 34–69%): it uses the hand check that makes the observed rate accurate, and the legal-name recall set is the closest match to how we searched the registers. The fully automatic reading gives 42.0% (29–64%). An earlier draft divided the hand-cleaned rate by the automatic recall (51.5%); that mixes two ways of counting and overstates the share.
One bias we can’t remove. The recall was measured on map-listed businesses, which picked a web address, keep an online profile and often use their trade name in their domain. Registry businesses are probably harder to find by name, so the true share is more likely above 47% than below it.
Found-by-name rates (floor, unweighted within group). License groups: first four rows (n = 27, 39, 26, 295); regions: last five rows (n = 59, 59, 98, 80, 99).
Source: Registry sample; DBPR license file (Aug 10, 2026 copy) and Sunbiz (Oct 2026).
Contractors with a DBPR license were 1.7 times as likely to have a findable site as businesses with no state license, +14 points (95% CI +4 to +24). Age makes little difference: businesses registered in the last three years were findable 26% of the time, those 20 years or older 22%.
The basics are solved; the self-service rung is empty
Once a pool business has a website, the technical floor is in place. Every working site we crawled loaded over https with a certificate the browser accepted (on the map-list sites, 1 of 114 had a certificate problem). The drop-off comes with every step that asks the owner to do something.
Share of the 75 working websites of registered pool businesses. Self-service rows are hand-checked floors.
Source: Ruberio crawl of registry-found sites, Oct 30, 2026.
Share of working registry sites that meet this rung and every rung above it.
Source: Ruberio crawl, Oct 30, 2026.
Read in order, the ladder thins quickly: 41% of sites clear the own-domain email rung, 24% get as far as a linked social profile and 20% reach “kept current”. Only 2 of the 75 registry sites clear all six rungs.
Across all 186 sites, booking links went to Calendly, Housecall Pro, Setmore, Wix Bookings, 1 own scheduler; logins to MyServiceAccount, Pool Brain, Service Fusion, ServiceTitan, 2 own portals; payment to QuickBooks, Wix pricing plans. A further 3 sites embed a quote-request form (Jobber), which we don’t count as self-service. If an owner’s route software already has a booking page or customer portal, linking it from the website is a small step; we can’t tell from outside how many owners have one unlinked.
Per 100 registered pool businesses
Combining the website rate with what those websites show gives a rough picture for the whole registered industry. This is a model: it assumes the sites we could find look like the ones we couldn’t.
| Of every 100 registered businesses… | Floor | Estimate | Range |
|---|---|---|---|
| Have a website (all of them secure) | 22 | 47 | 34–69 |
| Have schema.org markup | 15 | 33 | 24–48 |
| Link a social profile | 15 | 33 | 24–48 |
| Show own-domain email | 9 | 20 | 14–28 |
| Have any analytics tag | 7 | 16 | 12–24 |
| Let customers book, log in or pay online | 0.9 | 1.9 | 1.4–2.8 |
Floor = sites we found (21.6%) × share with the feature. Estimate and range use the main corrected website rate (47%, 34–69%). The range does not include uncertainty in the site-level shares, which for the last row is large (95% CI 1–11% of sites).
Register sites and map-listing sites sit on the same ladder
A list built from map listings is the obvious place to start; we started from the state registers instead and ran the same crawler on both. In the four regions the map list covers, 4 of the 69 registry-found sites (5.8%) were also on the map list. That small overlap says little on its own: those regions hold roughly 2,584–5,686 registered pool businesses with a website, against 167 map-list sites, so even if every map-list business were in the register the overlap could not exceed 2.9%–6.5%. The useful question is whether the two lists look different. Mostly, they don’t.
Six-rung score (see Method). Mean 3.49 vs 3.54.
Source: Ruberio crawl, Oct 30, 2026.
The average score differs by −0.05 points (95% CI −0.40 to +0.28). Self-service is rare in both: 4.0% of registry sites and 10.5% of map-list sites, a difference of −7 (−14 to +2) points that our sample can’t distinguish from zero. The clear differences (bold below) are about marketing and contact details, not plumbing: registry-found sites less often carry an analytics tag or link a blog, and more often show an email address, also one on their own domain.
| Feature | Registry sites | Map-list sites | Difference, points (95% CI) |
|---|---|---|---|
| Secure (https, valid certificate) | 100% | 96% | +4 (−1 to +10) |
| Shows an email address | 83% | 68% | +15 (+2 to +27) |
| Email link on own domain | 41% | 27% | +14 (0 to +28) |
| Has a contact form | 85% | 74% | +12 (0 to +22) |
| Any schema.org markup | 69% | 75% | −6 (−19 to +7) |
| LocalBusiness schema | 41% | 53% | −11 (−25 to +3) |
| Links a social profile | 69% | 76% | −7 (−20 to +6) |
| Copyright 2025 or 2026 | 65% | 69% | −4 (−18 to +9) |
| Copyright 2024 or older | 15% | 18% | −4 (−14 to +8) |
| Any analytics tag | 35% | 52% | −17 (−30 to −3) |
| Links a blog or news page | 19% | 36% | −17 (−29 to −4) |
| Chat widget | 11% | 14% | −3 (−13 to +7) |
| Book, log in or pay online | 4% | 11% | −7 (−14 to +2) |
Registry n = 75, map list n = 114 working sites; 4 businesses are in both lists. Bold = the 95% interval excludes zero. With these sample sizes, differences under about 10 points cannot be told apart from noise.
What’s under the hood: builders, old footers, quiet blogs
Platform detected from page code, n = 75. “Other builders”: Hostinger, Shopify, Weebly, Webflow, Google Sites and similar.
Source: Ruberio crawl, Oct 30, 2026.
Footers. 43 of 75 registry sites show 2026, 6 show 2025, 11 show 2024 or older (oldest: 2015), and 15 have no usable year at all (missing, or a year in the future). A footer with an old year tells a visitor nobody has looked at the site in years, whether or not that’s true.
Blogs. 14 sites link a blog or news page; on 5 of them we could read a date. The newest post was 2 days, 301 days, 339 days, 582 days, 899 days old. A dated, abandoned blog is the single easiest stale signal to remove.
Forwarding. 12% of the registered businesses’ domains forward to a different web address, usually a second brand or a newer domain.
By region
Region makes no visible difference to how mature a site is, though the regional samples of working sites are small.
| Region | Working sites | Mean score (0–6) | Website found (floor) |
|---|---|---|---|
| Tampa Bay | 13 | 3.7 | 25% |
| Orlando | 10 | 3.5 | 22% |
| South Florida | 18 | 3.4 | 20% |
| Southwest | 20 | 3.6 | 26% |
| Rest of Florida | 14 | 3.1 | 17% |
Licensed contractors’ sites score 3.6 on average (n = 25), unlicensed businesses’ sites 3.4 (n = 50): the license predicts whether there is a site, not how good it is. Too few sites per license class to split further.
Limitations
- The website rate rests on an assumption. The correction assumes our name search misses registry businesses’ sites as often as it missed map-listed ones; if it misses more, the true share is higher. Like-for-like readings range from 41% to 47%; the interval (34–69%) covers random error only. The recall set is small (55 businesses). A search with Google would settle it, but we used only sources without fees or accounts, and a web-search calibration we tried could not find even known sites.
- The list is not every pool business. Cleaning-only businesses without a pool word in their name, and sole proprietors who never registered a name, are missing. If those businesses are less online than the rest, the industry as a whole is less mature than shown.
- Registered isn’t trading. Dormant LLCs are part of the list and have no website, so the rate describes registered entities. They don’t affect site-level figures.
- Site-level figures depend on findable sites. Sites whose address resembles the business name may differ from those that don’t. Bot-blocked sites (10%) are excluded.
- Rule-based extraction. Call, CTA, price and license detection were hand-validated at 97–100% agreement. Every self-service positive was checked by hand, and 30 random negatives showed no misses; the crawler’s authors estimate about 90% recall for tools with a recognizable vendor and less for widgets that only appear behind a button. Self-service counts are floors. Social and schema detection were not separately validated; email domain type is only known when the address is a link.
- Small numbers at the top. 3 self-service sites is a small count: the 95% interval for the registry rate runs from 1–11%. The conclusion “rare” is firm; the exact rate is not.
- Data age. License data date from Aug 10, 2026; the crawl is a single day (Oct 30, 2026).
What an owner can do with this
A ten-minute digital health check. Do it on your phone, on mobile data, the way a new customer would. Each “yes” on the six scored items is one rung. Among the 75 registered pool businesses’ sites we could find and read, the average was 3.5 out of 6, and 21% scored 5 or more.
The six rungs (score yourself)
- Secure. Type your web address without “https”. Does it land on https with a padlock and no warning?
- Own-domain email. Is the email on your site an address at your own domain, not a Gmail or Yahoo address? Ask whoever holds your domain what an address or forwarding at it would take.
- Machine-readable. Paste your address into Google’s Rich Results Test. Does it find LocalBusiness data with your name, address and phone? Only 41% of sites have it.
- Connected. Do your social profiles link from your site, and are they the ones you still post on?
- Kept current. Does the footer say 2026? Is there a blog or “news” page whose newest post is more than a year old? Post something or remove the link.
- Self-service. Can a customer do one thing without calling you: book a visit, log in to see visit reports, or pay an invoice? If your route software has a booking page, customer portal or payment link, link it from your menu.
Three more checks
- Can people guess your address? If someone only has the name on your truck or your state registration, could they type your web address? Our name search found less than half of known sites.
- Do you know your visitors? Only 35% of registry sites have any analytics tag. Without one you can’t tell whether the site is used at all.
- Does everything still work? Open every page on your phone. Submit your own contact form and check the message arrives.
Rungs reached: ___ of 6
Average of registry sites we could read: 3.5
First fix this week: ______________________
Sources
- Florida Department of Business and Professional Regulation, Construction Industry Licensing Board licensee extract (CONSTRUCTIONLICENSE_1.csv), public records download, via the archive.org copy of Aug 10, 2026. https://www2.myfloridalicense.com/sto/file_download/extracts/CONSTRUCTIONLICENSE_1.csv — copy retrieved Oct 30, 2026.
- Florida Department of State, Division of Corporations (Sunbiz), quarterly corporate and fictitious-name data files (corporate data Oct 7, 2026; fictitious names Oct 3, 2026), sftp.floridados.gov — accessed Oct 27, 2026.
- The 2026 Florida Statutes, § 489.105 Definitions (pool/spa contractor classes and the cleaning exemption). http://www.leg.state.fl.us/statutes/ — accessed Oct 30, 2026.
- Sarasota County Health Department, pool contractor licensing information (2016), and DBPR CILB form 5-L, used to interpret license classes A, B and C — accessed Oct 30, 2026.
- U.S. Census Bureau, 2020 ZCTA to County relationship file. https://www2.census.gov/geo/docs/maps-data/data/rel2020/zcta520/ — accessed Oct 30, 2026.
- Schema.org, LocalBusiness type. https://schema.org/LocalBusiness — accessed Oct 30, 2026.
- Google Search Central, Local Business (LocalBusiness) structured data. https://developers.google.com/search/docs/appearance/structured-data/local-business — accessed Oct 30, 2026.
- Ruberio crawl of pool business websites, Florida, Oct 30, 2026 (feature extraction and validation notes available on request).
- Map-listing export of Florida pool businesses, Oct 27, 2026 (website, region and category only; no ratings or reviews used).
About this research
Who made this
Ruberio is a Dutch design studio: Ruben designs, Sven answers every email. We’re focused on websites for pool service companies and fence contractors in Florida. We ran this study because we wanted to know how many pool businesses have a website at all, and what is on them, before telling anyone what theirs should do.
The data and scripts are available on request; we publish only totals and never name a business. If you spot an error, email hello@ruberio.com and we’ll correct it and note the change.
Cite this study
Ruberio Research (2026). Only about 8 in 100 Florida pool service websites let a customer book, log in or pay online (No. 03). Data as of Oct 30, 2026. us.ruberio.com/research/pool-service-digital-maturity-florida/
Questions
Why is the website share an estimate and not a count?
We could find a site for 22% of the 395 sampled businesses by searching for their name as a domain, and that method finds only 42%–47% of sites we already know exist. The 47% corrects for that. Like-for-like readings range from 41% to 47%; if our search misses more often than we assumed, the true share is higher.
Does “8 in 100” mean the rest have no online booking at all?
Not necessarily. Tools that only appear behind a button can be missed, so the self-service counts are floors. The conclusion that it is rare is firm; the exact rate is not.
Who is missing from the register list?
Routine pool cleaning needs no state license in Florida, so cleaning-only businesses appear only through Sunbiz, and only if their name contains a pool word. Sole proprietors who never filed a name are missing too. Dormant LLCs stay in the list, so the website rate describes registered entities, not only businesses that are trading.
