Everyone in contractor marketing will tell you the competition is weak. We went and measured it instead. In October 2026 we checked 51 small, independent US contractor sites — roofing, plumbing, HVAC, electrical and landscaping — on the things that decide whether a homeowner stays and whether a search engine can understand the business at all. Every figure below is ours, with the sample size attached. No email required.
Small, independent US contractor businesses drawn from public listings via OpenStreetMap and Overpass, across five trades, filtered to those with a working website. Independents, not national franchises — a firm with its own marketing department would flatter every figure here.
Speed from Google's own Lighthouse on a throttled mobile connection. Structure by fetching and parsing each page: schema, titles, headings, sitemap, certificate. Nothing is estimated and nothing is modelled.
A sample of 51 is enough to show a pattern and not enough to publish a national average. Sample sizes differ per check because not every check applies to every site, so each finding carries its own count rather than hiding behind one tidy number.
Measured on a throttled mobile connection, the way most people arrive.
The median, not the worst case — the worst was forty. A homeowner with a burst pipe does not wait nine seconds; they go back and call the next result. n = 50.
Google publishes 2.5 seconds as the line between good and poor. Half the sites took over ten. This is not a tail of bad sites, it is the normal condition of the industry. n = 50.
A quarter were over five megabytes. The heaviest single page we measured was forty-six — roughly a film trailer, delivered to someone standing in a driveway on cellular data. n = 50.
36% scored under 50, Google's red band. The bar being this low cuts both ways: it is why a fast site stands out, and why “we built you a website” means very little on its own. n = 50.
Google held real-user performance data for seven of the fifty. For the other forty-three, not enough people visit for Google to have measured anything at all. n = 50.
Structured data and page basics decide whether a machine can read the business at all.
Nearly half have never stated, in the one format built for stating it, what they do or where. A further quarter carry structured data a plugin added, which describes the website and says nothing about the business. n = 51.
Not one line of JSON-LD. To an answer engine deciding who to name for a trade in a city, those businesses are prose to be guessed at rather than a record to be read. n = 51.
Median title length was 60 characters and half ran past it. For a local business the end of the title is usually the town, so the part that makes it local is the part that disappears. n = 51.
Which asks Google to choose between several pages all claiming to be the same thing. It usually picks the home page, and the service pages never surface. n = 44.
Either none at all or several competing. A small signal by itself, and a reliable sign that nobody has looked at the page the way a search engine does. n = 51.
The cheapest thing on this list to fix. Counted strictly: a site only qualifies when /sitemap.xml is absent and robots.txt points at nothing that resolves. n = 51.
Two findings that are rare, and serious where they occur.
Three of these businesses send a homeowner's phone number over plain HTTP, and Chrome warns the visitor before they ever see the page. n = 51.
Four businesses appear in public listings with a website that no longer exists — three domains that do not resolve and one whose certificate fails. Every citation pointing there is a dead end, and the owner almost certainly does not know. n = 67.
The useful finding is not that contractor websites are slow. It is that almost all of them are slow in the same few ways, and none of those ways is hard to fix. A median of nine point eight seconds is not the result of a hard engineering problem — it is uncompressed photographs and plugins nobody audited.
The structural findings matter more than they look. Nearly half of these businesses have never told a search engine what they are in the format built for saying it, and a fifth have said nothing machine-readable at all. That was survivable when ten blue links were the whole game. It is not survivable when an answer engine is choosing one business to name.
Which is the opportunity, and we will say the obvious thing: we sell the work that fixes this. But the figures are the figures whoever does it. If you measure your own site against them and it already clears the bar, you do not need us — and you will know that for free rather than paying someone to tell you.
Measured and written by Cameron, founder of Local Leader.
Fieldwork October 2026 · last reviewed
We drew small, independent US contractor businesses from public business listings via OpenStreetMap and Overpass, across roofing, plumbing, HVAC, electrical and landscaping, then kept the ones with a working website. It is a sample of independents, not of the industry's largest firms — a national franchise with a marketing department is a different animal and would flatter the numbers.
Because not every check applies to every site. Speed was measured on the 50 sites that returned a usable mobile result; the structural checks ran on 51; the duplicate-title check needs more than one page to compare, so it ran on the 44 with enough pages. The count is stated on each finding rather than averaged into one tidy number that would be wrong.
No, and conflating the two inflates this kind of research. Plenty of contractor sites sit behind a firewall that refuses automated requests while serving a human perfectly well. We treated a site as broken only when the domain failed to resolve or the certificate failed, and checked robots.txt rather than assuming.
Yes. They are our own measurements and you are welcome to quote them with a link back to this page. If you want the methodology in more detail than is written here, ask and we will send it.
It means the bar is low, which is useful information either way. If your site loads in under 2.5 seconds on a phone and carries correct LocalBusiness schema, you are already ahead of most of the companies you compete with. If it does not, the fix is usually a few days of work rather than a rebuild.
Apply to work with us and find out if you're the right fit to lead your market.