Essay · 01

Most UK business websites are fine. We measured it.

Most of the businesses we checked were refused as prospects because there was nothing wrong with the site. The refusal is the point; the pages around this one carry the measurement.

Sam, Orbcrawl · 5 September 2026 · Figures are from the 5 September 2026 reading; see Evidence for later readings.

Dated note. This essay was written on 5 September 2026 from that day’s reading, while the crawl was still running. Later work revised how the figures are counted, so its shares should be read as of that date and not as current.

The claim

Many agencies, lead-list vendors and cold-email tools in this market start from the same premise: small businesses have bad websites, so find small businesses and sell them websites. We built a pipeline to test that premise across UK districts. For most businesses it is wrong: seven in ten sites we could measure had nothing worth writing to their owner about.

Illustrative, not measured values: seven in ten stop short of the line.

What “fine” means

It is not an opinion about design. The crawler fetches the homepage once and records only what the response can prove, through deterministic checks on what a visitor sees. A site that passes is not treated as a prospect at all — a tested rule in the code, not a cell in a spreadsheet. A site we could not read is recorded as unread, not as bad.

The judgement is in what counts as fine; the measurement underneath is not. Nothing here has yet been calibrated against real-world outcomes, and the same page fetched again gives the same result. The lookup shows what was observed at a domain.

Why the refusal matters

Lists are priced by length, so “your website needs replacing” goes to the café whose site is fine as readily as to the one that has not changed in fifteen years, and the first owner learns that the sender did not look. Refusing most of the market is what looking costs. We think it is also the hardest part to copy, because it is a decision rather than a dataset — what we are building.

The worst website is not the best prospect

A dead website scores worst. It is also, very often, a dead business: the site went when the lease did. The same score on a business that demonstrably traded this year means the opposite — identical site condition, opposite value.

Illustrative: same score, opposite verdict. What separates them is a reading of the business, not the page.

So the site score is one reading among several, and the rest are about the business, not the page — kept apart on purpose. The lookup shows what was observed at the domain and stops there.

What the refusal leaves standing

Most of the register is refused, for reasons that are about the site, the business or the law, and what remains is a small part of it. The arithmetic is its own essay.

Illustrative, not to scale: each stage keeps less than it discards.

What “fine” does not mean

Fine means no measurable technical problem on the homepage. It says nothing about bookings, or a contact form broken since a plugin update. That is a different question, and this measurement does not answer it — a known limit.

Try the lookup on your own domain. Most who do will be told there is nothing wrong. That is the point.