Most contractor websites are invisible to search engines by construction
We fetched every deck contractor website in our dataset and measured what came back. 72.7% of licensed deck-capable contractors have no website at all — and of the ones that do, 85.3% carry no LocalBusiness structured data and 65.7% have no analytics installed. This is the sequel to our licensing report, and it answers the question that one raised.
Key findings
- Website ownership splits four to one on where a contractor is listed. Deck specialists in a trade directory or manufacturer network are 18.4% without a website (640/3,471). Those appearing only in a licence registry are 71.7% (1,012/1,411). Averaging the two would describe neither, so we never do.
- The technical gap does not close. The directory group is four times more likely to have a website — yet 66.8% of their live sites carry no LocalBusiness schema, against 73.1% for the registry group. Marketing sophistication buys a website; it does not appear to buy a findable one.
- 65.7% of the wider trade’s live sites have no analytics of any kind. No Google Analytics, no tag manager, no pixel (1,218/1,853). Those owners have no way of knowing whether their website has ever produced a job.
- Broken is a different problem from absent. 10.5% of fetched specialist sites did not load, 1.6% served an expired or invalid TLS certificate that browsers block outright, and 6.1% of the wider trade’s domains are parked or for sale. These firms were online once.
The three populations
Reported separately throughout, because they behave differently. Percentages appear next to their numerator and denominator; the third population is a seeded random sample and its audited figures carry a 95% Wilson confidence interval.
| Population | Businesses | Sites fetched | No website |
|---|---|---|---|
| Deck specialists — trade directory / manufacturer network | 3,471 | 2,831 | 18.4% |
| Deck specialists — licence registry only | 1,411 | 399 | 71.7% |
| All licensed deck-capable contractors (sample) | 408,555 | 2,000 | 72.7% |
What the live websites are missing
Across all 1,853 sampled pages we received and read from the wider licensed trade. Pages that failed to load, and pages served a bot challenge, are excluded from these denominators.
| Measured on the page | Share | Count |
|---|---|---|
| No LocalBusiness structured data | 85.3% | 1,581/1,853±CI 83.6–86.9% |
| No analytics or tracking installed | 65.7% | 1,218/1,853±CI 63.5–67.9% |
| No social profile linked | 50.7% | 940/1,853±CI 48.5–53% |
| No structured data of any kind | 47.7% | 884/1,853±CI 45.4–50% |
| No Open Graph tags — shares render as a bare link | 35.7% | 662/1,853±CI 33.6–37.9% |
| No h1 heading | 28.3% | 524/1,853±CI 26.3–30.4% |
| No phone link, email link or contact form | 22.2% | 411/1,853±CI 20.3–24.1% |
| Copyright notice over two years old | 15.6% | 289/1,853±CI 14–17.3% |
| Not mobile-friendly — no viewport tag | 8.9% | 165/1,853±CI 7.7–10.3% |
| Served without HTTPS | 8.4% | 155/1,853±CI 7.2–9.7% |
| Takes over 5 seconds to load | 1% | 18/1,853±CI 0.6–1.5% |
No website, by licensing jurisdiction
Deck specialists found in a licence registry only — the population that has a jurisdiction attached. Jurisdictions with fewer than 25 such businesses are omitted, because there the percentage moves on a single firm. This is where the licence was issued, not necessarily where the firm works.
| Jurisdiction | Deck specialists | No website | Share |
|---|---|---|---|
| AZ | 94 | 76 | 80.9% |
| CT | 51 | 40 | 78.4% |
| OR | 113 | 85 | 75.2% |
| WA | 398 | 298 | 74.9% |
| FL | 257 | 190 | 73.9% |
| DE | 44 | 32 | 72.7% |
| CA | 382 | 255 | 66.8% |
| IA | 36 | 24 | 66.7% |
The spread runs from 80.9% in AZ to 66.7% in IA. That is a narrower band than the four-to-one gap between directory-listed and registry-only businesses, which is the point: where a contractor is listed predicts web presence far better than which state licensed them.
Does a longer-established firm have a better web presence?
Slightly worse, and by a modest margin. Licence issue dates exist only on registry rows, so this sits inside that population and should be read against its 71.7% base rate rather than any blended figure.
| Years since licence issued | Businesses | No website |
|---|---|---|
| 0-5 years | 604 | 70.2% |
| 5-10 years | 329 | 73.3% |
| 10-20 years | 251 | 76.1% |
| 20+ years | 204 | 74.5% |
The gradient from 70.2% at 0-5 years to 76.1% at 10-20 years is real but small, and it does not continue into the oldest bucket. We would not describe this as a trend; a few points across four buckets of a few hundred businesses each is a hint, not a finding, and we would rather say so than dress it up.
- Every live-site figure is render-verified. A single HTTP GET cannot distinguish a React site from a derelict one — both return a nearly empty document. Re-checking in a headless browser overturned 179 of 452 “poor” verdicts and 119 of 441 “dead” ones, so raw HTTP percentages are not published.
- Bot challenges are excluded, not counted as failures. A page served a challenge was never received, and cannot be judged in either direction.
- Specialist figures describe this dataset, not every deck business in the United States. Wider-trade figures cover nine licensing jurisdictions and are never national; several large states license no residential contractors at state level at all.
- The wider-trade sample is a seeded random draw, so the same sample can be reproduced and any figure here independently re-checked.
- “No website” means none in the record we hold. A firm trading through a social page alone counts as having none, and a website never filed with its licensing authority is undercounted.
Frequently asked questions
What share of deck contractors have a website?
It depends entirely on which deck contractors you mean, and that is the most important finding in this report. Among businesses listed in a trade directory or a manufacturer's contractor network — NADRA members, Fiberon programme contractors — 18.4% have no website (640 of 3,471). Among deck specialists that appear only in a state licence registry, the figure is 71.7% (1,012 of 1,411), and across all 408,555 licensed deck-capable contractors in the nine jurisdictions we hold it is 72.7%. That is a roughly fourfold gap between the first group and the other two. It is not evidence that joining a directory causes a business to build a website; the causation almost certainly runs the other way, and both are downstream of a firm that has decided to market itself. The practical consequence is that any single "X% of contractors have a website" figure is meaningless unless it says where the contractors were found. We publish the three separately for that reason and never average them.
Why does LocalBusiness schema matter for a contractor?
Schema.org markup is how a page tells a search engine what kind of thing it describes. A LocalBusiness or HomeAndConstructionBusiness node states, in machine-readable form, that this page represents a real trading business with a name, an address, a service area and opening hours. Without it a search engine has to infer all of that from prose, and increasingly declines to — local rich results, map placements and AI-generated answers lean heavily on structured data because it is unambiguous. The measured adoption is strikingly low. Among directory-listed deck specialists — the most marketing-forward group in this dataset — 66.8% of live sites carry no LocalBusiness schema (1,496 of 2,238). Across the wider licensed trade it is 85.3% (95% confidence interval 83.6–86.9%). So the great majority of contractor websites that do exist are, to a search engine, an unlabelled document that happens to mention a town.
Does having a website mean a contractor is findable online?
The two are much further apart than they look, and the gap does not close with marketing sophistication. Directory-listed deck specialists are about four times more likely than registry-only ones to have a website at all — 18.4% without, against 71.7%. But on the technical signals that decide whether that website can be found, the two groups are nearly identical: 66.8% and 73.1% respectively have no LocalBusiness schema. Analytics tells the same story from a different angle — 41.9% of directory-listed contractors' sites have no analytics or tracking of any kind installed, meaning the owner has no way of knowing whether the site produces work. Building a website and building a website that functions as a channel are separate projects, and the data suggests the trade has largely done the first and not the second.
How many contractor websites are broken rather than absent?
A meaningful minority, and they are worth separating from the ones that were never built. Of the 2,831 directory-listed deck specialists whose website we fetched, 10.5% did not load at all (298 sites), and a further 1.6% were reachable but served an expired or invalid TLS certificate, which every current browser blocks with a full-page warning before the visitor sees anything. Across the wider licensed trade, 6.1% of fetched domains were parked or listed for sale — a business that had a domain, stopped paying for the site, and now points customers at a registrar's advertising page. Another 22.2% of live sites carry no phone link, no email link and no contact form anywhere on the home page. These are not firms that never got online; they are firms whose online presence has quietly stopped working, usually without anyone telling them.
How was this measured, and what are the limits?
Every website in the dataset was fetched with a single HTTP GET, following redirects, and each signal in this report is a property of the document that came back. Any verdict that rested on a page appearing empty, and every site that failed to load, was then re-checked in a real headless browser, because a React or Next.js site returns an almost empty HTML shell and scores identically to a derelict one. That second pass overturned 179 of 452 "poor" verdicts and 119 of 441 "dead" ones, so single-pass HTTP figures are not published here. Pages served a bot challenge are counted separately and excluded from on-page percentages, since a page never received cannot be judged in either direction. The limits: our specialist figures describe the deck businesses in this dataset, not every deck business in the United States; the wider-trade figures cover nine licensing jurisdictions and are never national; `state` means the licensing jurisdiction, not where a firm works; and New York here is New York City's DCWP, not New York State.
Can I reuse these figures?
Yes. Quote or reproduce anything on this page with attribution to DeckMath and a link back. Please cite the snapshot date — September 2026 — because websites change continuously and a re-run in six months will produce different numbers. Two requests, both about not misrepresenting the data. First, do not blend the three populations into a single "contractors" figure: the directory-listed group and the registry-only group differ by a factor of four on website ownership, and an average of the two describes neither. Second, do not present the wider-trade figures as national, because they cover nine licensing jurisdictions rather than fifty, and several large states license no residential contractors at state level at all. The wider-trade sample is a seeded random draw, so the same sample can be reproduced from the database and any figure here re-checked.