We scanned 73 contractor websites in Palm Beach County. Here is what we found.

VoltFound Learn · Updated 2026-09-12 · by AppVolt LLC

Between 9 and 12 September 2026 we ran our scanner over the public websites of 81 local contractor businesses in Palm Beach County and just south of it — plumbers, electricians, roofers, HVAC companies, pool services, pest control, landscapers, painters, garage-door and handyman businesses in West Palm Beach, Palm Beach Gardens, Lake Worth, Jupiter, Wellington, Deerfield Beach and Pompano Beach.

This is what we found. No business is named anywhere in this article, and this is not a ranking or a scoreboard. It is a description of a pattern, published so that any owner can check the same things on their own site for nothing.

How we did it

• We read each site the way a search crawler does: the raw HTML of the homepage, the robots.txt, the sitemap, and up to 45 pages per site.
• Eight sites are excluded from every number below: three blocked our reader outright (bot protection — that is not a verdict on the site) and five did not respond at all. That leaves 73 readable sites.
• For internal links we stripped the navigation, header, footer and sidebar blocks, and additionally treated any link that appears on more than 80% of a site's pages as menu furniture. What is left is links inside the actual body text.
• Where we mention an AI assistant, we asked it a plain customer question and recorded what it said at that moment. Nothing here is a prediction.

Finding 1: most of these sites are technically fine

The median overall score was 91 out of 100. Forty-eight of the 73 scored 90 or above. The lowest was 57, the highest 98.

If you own one of these businesses and you have been told your website is “broken” for search, the data does not support that. The basics — a title, a description, HTTPS, a readable page — are mostly in place.

Finding 2: the things that do fail are small and repeat

Counting the checks that actually failed across the 73 sites:

• Homepage images with little or no alt text — 26 sites (36%)
• Resources loaded over plain http on an https page — 9 sites (12%)
• No H1 heading on the homepage — 7 sites (10%)
• No meta description — 7 sites (10%)
• Very little readable text on the homepage — 3 sites (4%)
• No title tag in the raw HTML — 1 site
• Canonical tag pointing at a different site — 1 site

Alt text is the one to start with, because it is the cheapest. A picture with no description is a blank space to anything reading the page without eyes — a search crawler, an AI assistant, or a customer using a screen reader.

Sixteen of the 73 homepages had under 500 words of readable text. The median was a little over a thousand.

Finding 3: the pages do not talk to each other

This is the pattern that surprised us, and the score above does not measure it.

On a ten-site subset we looked for groups of pages on the same site aiming at the same service in the same town — a service page, a city page and a blog post all about water heater repair, for instance. Six of the ten sites had at least one such group; 14 groups in all, holding 56 pages between them. Set aside the 14 pages an owner would pick as the one to send customers to, and 42 remain. Forty-one of those 42 had no link, inside the body text, to the target page.

All 42 did link to it — through the top menu, the same menu that sits on every page of the site. That is the part worth sitting with: the connection exists, but it is the kind every page has to every other page, so it tells a machine nothing about which one matters.

Seven of those ten sites had half a body-text internal link per page or fewer across the whole site; three had none at all. We repeated the measurement on 14 more sites and found the same pattern.

Why it matters: when several of your own pages say the same thing about the same service in the same town, a search engine or an AI reader has to guess which one is the real page for it. It does not always guess the one you would pick. Choosing for it costs nothing but your time.

Finding 4: a high score does not mean an assistant knows you exist

On 9 and 10 September we also asked an AI assistant a plain customer question for nine of these businesses — “who are the best [trade] in [city]?” — making sure each was asked about its own trade and its own town.

On that day, none of the nine was named. The assistant named other businesses instead, three to six of them each time. Those nine sites had a median score of 96 out of 100, and one of them has thousands of reviews at 4.9 stars.

Then we re-ran part of the check on 12 September, and one of the nine was named — first in the list. We are reporting that because it is the most important caveat in this article: these answers move. A single result is a snapshot, not a scoreboard, and nobody can promise a mention. Anyone who tells you otherwise is selling something we would not sell.

What the snapshot does tell you is that a clean technical score and a wall of five-star reviews are not, on their own, enough to be legible to an assistant. Those are different questions, and it is worth knowing the answer to both.

Five things you can check on your own site today, for nothing

1. Google your own business name and read the blue headline. That line comes from your website, not from Google — it is the title tag. If it says “Home”, rewrite it as what you do, the town you do it in, then your business name, in roughly sixty characters.
2. Open your homepage and count the pictures with no description. Every site editor has an alt-text or “image description” field. Fill it with what the picture actually shows.
3. Search your own domain plus a service in quotes — for example yourdomain.com "water heater repair". If several of your own pages come back, pick one as the target and add a plain text link to it, inside a real sentence, on each of the others.
4. Open yourdomain.com/robots.txt. If it blocks the AI search crawlers, you have opted out of being read for those answers.
5. Ask an assistant the question your customer would ask and write down who it names. Do it again in a month. Look at the trend, not one answer.

Why we published this

We build a scanner, so we have an obvious interest here. Two things we want to be straight about. First, the data says most local contractor websites are not a disaster, which is not the story a tool vendor usually tells. Second, our own score did not predict whether an assistant named a business — and we would rather say that out loud than have you find out after paying us.

If you want the same report for your own site, the scan is free, takes about a minute and needs no signup.