636 businesses with working websites that nobody can find

Every business in the Local Digital Visibility Index has a website. That was a condition of being measured at all. We check each one against twelve keywords its own customers would plausibly search — the sector and the town, the service and the town, the obvious variations — and record where it appears in the first hundred results.

636 of 1371 appear nowhere in any of them. Not ranked badly. Not on page four. Absent from the first hundred results for all twelve.

605 of those are also never named by an AI assistant asked to recommend a firm in their sector. They have a website, and as far as either channel is concerned they do not exist.

Every figure here is computed from the published datasets when this page is built.

It is not the websites

The intuitive explanation is that these are bad websites — slow, broken, thin, built in 2011 and left alone. That would be a comfortable finding, because it comes with an obvious remedy.

It is not what the data says.

Pillar	The 605 invisible	All 1371 measured
Speed & CWV	65	64
Technical	75	88
Content & trust	35	49
Local presence	68	76

On three of the four, the businesses nobody can find score the same as everybody else. Their sites load at the same speed. They are built to the same technical standard. 390 of the 605 score 70 or better on the Technical pillar — genuinely sound sites, properly served over HTTPS, mobile-friendly, indexable, with a valid sitemap.

They are competent websites that no one arrives at.

How it varies by sector

Businesses returning no ranking anywhere in the first 100 results for any of their sector's twelve keywords, as a share of each index.

The spread matters more than the average. Being invisible is not evenly distributed: it clusters by sector and town, which is what you would expect if the cause were competitive rather than technical.

In the crowded indices, a firm with a decent site is fighting directories, national chains and franchise location pages for a small number of slots. In the thinner ones, the same firm would rank. Nothing about the website changed between those two situations.

What is actually happening

Three explanations fit this data, and they are not exclusive.

The keyword basket is contested by people who are not local businesses. In several sectors, most of the first page belongs to directories, comparison sites and national brands. A twelve-keyword basket aimed at “accountants in Gloucester” is largely a competition against Yell and a handful of national firms, not against the other accountants in Gloucester.

A technically sound site is table stakes, not an advantage. When the median business scores 88 on Technical, doing the technical work well moves you to the middle of the pack. It is necessary and it is not differentiating — which is precisely why the pillar we weight most heavily turns out to predict findability so poorly.

There is nothing on most of these sites to rank. The median Content & trust score across the whole index is 49. That pillar asks undemanding questions — is there an about page, are credentials stated, is there evidence anyone has updated this. A median of 49 out of 100 means most of these businesses have published almost nothing about what they do. A search engine cannot rank a page that was never written, and an assistant cannot recommend a firm it can read nothing about.

What this does not prove

Invisible in this basket is not invisible in general. We test twelve keywords per sector. A firm that gets all its work through a specific niche term we did not test, or through referral, or from a Google Business Profile that performs well in the map pack, may be perfectly healthy and still score zero here. Ranking for nothing in our basket is not the same as having no customers, and no figure on this page should be read as a claim about anyone’s revenue.

One measurement, one quarter. Rankings move. A business absent on the measurement date may be present next month.

The AI half is one assistant. The 605 figure uses Perplexity only, on five prompts — a limitation our methodology page now states plainly after previously overstating it.

If this is your business

The useful question is not “is my site good enough” — for most of the firms above, it demonstrably is. It is “what have I published that anyone could find, and is it about the thing my customers actually search for?”

Concretely, in the order the data supports:

Check whether you rank for anything at all. Not your business name — anyone ranks for their own name. The service and the town.
Look at who does hold those results. If it is four directories and a national chain, the realistic target is a narrower, more specific query, not the head term.
Write the pages that answer real questions, and link to them from your homepage. The firms in this index that get found are, more than anything else, the ones with something published.
Make sure your Google Business Profile is complete. It is the one channel in this data where the invisible firms score measurably lower than everyone else.

Where your own firm sits

All 1371 measured businesses are listed on the indices, each with its own scorecard.

Every score above is recomputable from the open data using the open-source pipeline. If you think we have measured your business wrongly, each scorecard carries a correction link, and corrections are logged publicly with a date.