Finding

Of 90 UK business websites Whito measured on 16 November 2026, 27 had an llms.txt file for AI tools, but only 4 named an AI crawler in robots.txt and 1 blocked one.

Source: Whito, whito.co.uk/research/uk-business-website-census, 2026-11-16. Free to quote with a link to this page.

Last Updated on November 18, 2026

Short answer: Of 90 UK business websites Whito measured on 16 November 2026, 27 had an llms.txt file for AI tools, but only 4 named an AI crawler in robots.txt and 1 blocked one. The basics are almost universal: all 90 used a secure address and 88 were built for phones. What is missing is the detail people and machines use to trust a business: 33 showed a price, 56 had a phone number you can tap, and 52 said in machine-readable form what kind of business they are.

This is wave 1 of the Whito UK Business Website Census. The same checks will be run again on a bigger sample, so the figures can be tracked over time. The full data file is free to download below.

What we tested

We took the 1,167 UK business websites listed in the Whito directory on 16 November 2026 and picked 100 at random. 52 belong to local firms with a town on their listing. 48 belong to national brands. We opened each homepage once in a desktop Chrome browser and ran one script that asked the same questions of every site.

OutcomeWebsites
Loaded and measured90
Loaded, but could not be measured7
Did not load on two tries3
Total tested100

Of the 7 we could not measure, 2 showed a bot-check screen first and 5 would not let our test script run. We did not try to get round either. Of the 3 that did not load, 2 had a domain name that no longer exists. Every share below is out of the 90 measured websites.

The basics are nearly universal

CheckWebsites passingShare
Secure address (https)90 of 90100%
Built for phones (viewport tag)88 of 9098%
Page language set87 of 9097%
Has a robots.txt file87 of 9097%
Has a meta description82 of 9091%
Links to a sitemap in robots.txt81 of 9090%
Every image has alt text set67 of 9074%
Exactly one main heading (H1)65 of 9072%

The two weakest basics are both about clarity. 25 homepages had no main heading or more than one. 23 had at least one image with no alt text at all, which screen readers need.

See how your own website compares on these checks with the free Whito Business Scorecard.

Trust details are where sites fall short

On the homepageAll 90Local firms (47)National brands (43)
Phone number you can tap to call56 (62%)34 (72%)22 (51%)
Email link47 (52%)31 (66%)16 (37%)
A UK postcode49 (54%)30 (64%)19 (44%)
At least one price in pounds33 (37%)13 (28%)20 (47%)
Names a review platform39 (43%)12 (26%)27 (63%)
Link to a privacy policy67 (74%)28 (60%)39 (91%)

Local firms are easier to contact. National brands are better at prices, reviews and legal pages. Two in five local firms had no privacy policy link on the homepage. The two groups are small, so treat gaps of less than about 20 points as noise.

We also looked for a company registration number in the homepage text and found one on 18 of 90. Our search pattern will have missed some wordings, so read that as “at least 18”. Whito has a fuller count in company numbers on UK trade websites.

Can machines tell what the business is?

CheckWebsitesShare
Any structured data (JSON-LD)71 of 9079%
Structured data that names the business as an organisation or local business52 of 9058%
Structured data using a local business type22 of 9024%
Main text only appears after scripts run2 of 902%

Structured data is a block of code that tells search engines and AI tools plain facts: this is a roofer, this is the address, these are the opening hours. Four in five sites had some. Fewer than three in five used it to say what the business is. Among local firms it was 24 of 47.

AI tools: 3 in 10 have an llms.txt file, almost nobody blocks crawlers

CheckAll 90Local firms (47)National brands (43)
Has an llms.txt file27 (30%)14 (30%)13 (30%)
robots.txt names any AI crawler4 (4%)22
robots.txt blocks an AI crawler1 (1%)10

An llms.txt file is a short text file that tells AI tools what a site is about. We opened four of the 27 by hand to check they were real files and not error pages. All four were. Some website builders now create this file on their own, so having one does not mean the business wrote it.

Earlier Whito checks used different samples, so they are not a like-for-like trend. For context: in June 2026 we found an llms.txt file on 18% of 49 sites in the AI door test, and in July 2026 92% of 120 independent sites had no AI crawler policy in the AI readiness study. This wave points the same way on crawlers: 86 of 90 say nothing about them.

Speed on one ordinary connection

MeasureResult
Middle time to first response0.27 seconds
Middle time to full load1.47 seconds
Took more than 3 seconds to load13 of 88 (15%)
First response slower than 0.8 seconds15 of 88 (17%)
Middle number of files loadedabout 72

These times come from one desktop computer on UK home broadband, each site opened once. They are not a lab test and not a phone test. Two sites were opened twice, so their times are left out. Local firms loaded slightly faster (middle 1.32 seconds) than national brands (1.59 seconds).

The Whito website basics score

To compare sites and track change, each measured site gets one point for each of ten checks: secure address, built for phones, meta description, exactly one main heading, page language set, alt text on every image, structured data naming the business, a tap-to-call number, a postcode or company number, and a privacy policy link.

Score out of 10Websites
1011
921
826
718
68
54
42

The average was 7.9 and the middle score was 8. Local firms averaged 7.7 and national brands 8.0. 14 of 90 scored 6 or lower. The score measures basics only. It says nothing about design, wording or whether the business is any good.

By sector and region

This wave is too small to compare most sectors or any regions, and we will not print numbers we cannot stand behind. Only home care had more than ten measured sites: 13 sites, average score 7.3, and 3 showed a price. Wave 2 will pick a set number of sites per sector and per region so those comparisons can be made.

How we did it

  • Sample frame: 1,167 different website addresses on published Whito directory listings, read on 16 November 2026.
  • Sample: the listings were put in a random order with a fixed seed and the first 100 were tested.
  • Test: each homepage opened once in desktop Chrome between about 16:55 and 17:15 UK time on 16 November 2026. One script ran the same checks on every site and also read its robots.txt and llms.txt files.
  • Local or national: a listing with a town is counted as a local firm. A listing with no town is a national brand.
  • Counting: every figure was counted twice, by two separate pieces of code, from the raw test output.

Limits

  • The Whito directory is not a random sample of UK businesses. It leans towards trades, home care and professional services, and every business in it has a website.
  • Only the homepage was checked. A price or company number on another page is not counted.
  • The checks are simple yes or no tests. “Shows a price” means at least one pound sign followed by a number. “Names a review platform” means the page mentions Trustpilot, Google reviews, Reviews.io or Feefo.
  • A missing alt attribute was counted. An empty one was not, because empty is correct for decorative images.
  • Speed is one visit from one place. It will differ on a phone or a slower line.
  • 10 of the 100 sites could not be measured, so the true picture may be slightly different.

Download the data

UK Business Website Census, wave 1 (CSV, 100 rows). One row per website with its sector, whether it is a local firm or national brand, and every check. Business names and web addresses are left out, because this is a count of the market, not a judgement on any one firm. Free to use with a link to this page.

Want to see how your own site does? Try the free Whito visibility check or the website legal details checker.

Common questions

How many UK business websites have an llms.txt file?

In Whito’s test of 90 UK business websites on 16 November 2026, 27 had one (30%). An llms.txt file is a short text file that tells AI tools what a website is about.

Do UK business websites block AI crawlers?

Almost none. Only 4 of 90 named an AI crawler in their robots.txt file and only 1 blocked one.

How many UK business websites show a price?

33 of 90 homepages (37%) showed at least one figure in pounds. National brands did so more often (20 of 43) than local firms (13 of 47).

How fast do UK business websites load?

On one desktop computer on UK home broadband, the middle website took 1.47 seconds to load fully. 13 of 88 took more than 3 seconds. This is one test on one day, not a lab result.

Is this a sample of all UK businesses?

No. The 100 websites were picked at random from the 1,167 business websites in the Whito directory. It leans towards trades, home care and professional services, so read it as a first wave, not a national figure.

Cite as: Whito, “UK Business Website Census, wave 1”, 16 November 2026.

Related reading
Corrections and right of replyThis page reports research that may name real businesses. Data is collected and checked on the dates stated. Company financials, insolvency and regulator action are taken from Companies House, company announcements or the regulator’s own notice.

If your business is named here and you think we have a fact wrong, email hello@whito.co.uk and we will check it and correct it. If you want to respond, send us a statement and we will publish it alongside. Full detail: Corrections, right of reply and removals.
Whito logo
Jacob Whitmore
Whito is an independent UK research publisher. We publish research on how UK businesses are found, trusted and priced: what they cost to market, how their sectors work, and who actually gets recommended. Studies are free to read, every figure carries the date it was checked, and nobody pays to appear.