← All resources

AI-invisible: we scanned 234 well-known UK SME brands twice. Not one scored 20/100.

11 min read

What you can do now

If you run a UK SME, do these four things now.

  1. Run the free scan Free. No credit card required. No sign-up.
  2. Check crawler access for the AI bots that matter (GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot, Google-Extended), and fix a blanket disallow if one is in place. Access is the first step, not a guarantee.
  3. Claim or update official listings you can action yourself (Businesses in ChatGPT / Business Hub, Google Business Profile, Bing Places, Apple Business), and keep your name, address and phone consistent.
  4. Publish or tidy obvious public signals: relevant schema.org types on the homepage, and /llms.txt if you want one. Google does not require llms.txt or a special schema type for AI Overviews or AI Mode.

Before a buyer, investor, lender or AI-powered search engine knows anything about your operations, it knows your public signals: your website, your technology stack, your job adverts, your Companies House filings, and your published policies. We ran Z3nith's 60-second public-signals scan across 234 well-known UK SME and mid-market brands, a balanced sample across retail, food and beverage, hospitality, manufacturing, legal, financial, professional services, healthcare, education, construction and logistics, to see what the outside world can tell about their AI readiness.

The short answer in the published June 2026 scan: almost nothing. Not one of the 234 brands scored 20 out of 100. The November 2026 public-signals scan of the same 234-brand published cohort also had none at 20. And that gap is now a commercial problem, not a cosmetic one.

What we measured

The public-signals scan reads only what anyone can see: the live website and its technology stack, careers pages and job advertisements, Companies House filing history, and published policies. It scores three of our seven readiness dimensions directly (Data Foundations, Technology & Infrastructure, Operations & Scale) and estimates the other four (Strategy, Talent, Governance, Investment) from external evidence alone. It is a deliberately conservative baseline: the same first impression a diligence team or an AI search engine forms.

We deliberately balanced the sample: up to 22 brands per sector, so the cross-sector comparison below is like-for-like rather than skewed by whichever sector we happened to scan most. The published cohort, unchanged for November, is the first 22 successful scans per sector, ordered by scan time and then by domain, except education and training (20) and wholesale and logistics (16).

What changed in November

Five months on, the same 234 brands are quieter, not louder. Average visible readiness fell from 10.2 to 8.9, only 2 brands now score 15 or more (6 in June), and none reached 20.

MeasureJune 2026 (published, n=234)November 2026 (n=234)
Mean visible readiness10.28.9
Median10.18.6
Range5.1 to 175.1 to 15.5
Bands 5 to 9.9 / 10 to 14.9 / 15 to 19.9 / 20 or above116 / 112 / 6 / 0166 / 66 / 2 / 0
Opportunity mean61 (published wording; underlying mean 60.8)43.9
Talent mean1.21.1
Strategy mean5.85.0
Governance mean5.24.4
Tech mean25.123.6
Investment mean28.916.7

Opportunity and investment draw on Companies House data, whose matching we tightened in October, so read those two rows with care.

Supporting counts from the same two scans:

  • Scored below 15: 228 of 234 in June, 232 of 234 in November.
  • Opportunity at or above 50: 169 of 234 in June, 81 of 234 in November.
  • Talent below 5: all 234 in both batches.
  • No strategy (strategy below 10): 178 of 234 in June, 182 of 234 in November. The June page described this as effectively no public evidence of an AI or technology strategy.
  • Governance below 10: 193 of 234 in June, 205 of 234 in November.
  • None reached 20 in either batch.

The headline numbers

Published June 2026 scan (n=234). This wording is the comparison already published for that batch:

  • Average externally visible AI-readiness score: 10.2 out of 100 (median 10.1; range 5.1 to 17). Roughly half the brands scored below 10; 228 of 234 scored below 15; none reached 20.
  • Talent signals were the quietest of all: every single one of the 234 brands scored below 5/100 (average 1.2). Careers pages and job adverts almost never mention data or AI skills, which is also what prospective hires read.
  • Strategy was near-invisible: 178 of 234 brands showed effectively no public evidence of an AI or technology strategy (dimension average 5.8/100). If leadership has a plan, the outside world cannot see it.
  • Governance barely registered (average 5.2/100; 193 of 234 below 10). Few publish anything beyond a standard privacy policy: no AI-use statement, no data-handling position a cautious B2B buyer or tender panel could point to.
  • Technology and investment were the strongest visible dimensions (25.1 and 28.9 average respectively). These brands clearly invest in their digital estate. They just say nothing about what comes next.
  • The opportunity gap is the story: the same companies averaged 61 out of 100 on AI opportunity, the firmographic upside if they adopted AI, with 169 of 234 scoring 50 or higher. Visible readiness of 10 against opportunity of 61 is the strongest case for action there is.

The sector league table

In the published June 2026 scan, with an equal sample per sector, externally visible readiness barely varies. Every June sector average sits within a 1.3-point band, but the opportunity scores show where the upside concentrates.

Every sector got quieter. Food and beverage fell furthest, to 6.9.

SectorJune brandsJune visible readiness (mean)June AI opportunity (mean)November brandsNovember visible readiness (mean)November AI opportunity (mean)
Construction & architecture229.564.5227.938.1
Financial services2210.863.12210.257.6
Healthcare & life sciences2210.663.0229.446.5
Retail & e-commerce2210.562.7228.642.3
Manufacturing2210.162.0229.250.5
Professional services2210.861.72210.445.5
Legal services229.860.8229.343.3
Education & training2010.059.7209.042.0
Wholesale & logistics169.559.7168.546.1
Hospitality & leisure2210.058.1228.136.4
Food & beverage2210.052.7226.935.2

Two patterns stand out in the June 2026 scan. Construction and architecture has the widest gap in the June 2026 scan: visible readiness of 9.5 against an AI opportunity of 64.5, a sector rich in schedule, cost and design data that its public presence barely hints at. Legal services is close behind (9.8 against 60.8): document-heavy, highly automatable work paired with one of the quietest public postures of any sector scanned. Across the board in that June scan, the readiness numbers cluster tightly while opportunity ranges from the low 50s to the mid 60s. The gap, not the level, is where each sector differs.

Why externally visible readiness now matters

It would be fair to object that a website is not the business. True, and it is exactly why we sell a questionnaire-verified assessment. But three audiences never get past your public signals:

  • B2B buyers and procurement teams increasingly ask AI-governance questions in tenders. A supplier with no visible position starts the conversation defending an information vacuum.
  • Investors and lenders read digital posture as a proxy for operational maturity. "AI-invisible" reads as "AI-unprepared", whether or not that is fair.
  • AI search engines, a fast-growing share of discovery, answer questions about your company from your public content. If you have not written it down, their answer is built from whatever else they can find.

What an SME should do about it

Start with the four steps a UK SME can do now:

  1. Run the free scan. Free. No credit card required. No sign-up.
  2. Check crawler access for GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot and Google-Extended, and fix a blanket disallow if one is in place. Access is the first step, not a guarantee.
  3. Claim or update official listings you can action yourself (Businesses in ChatGPT / Business Hub, Google Business Profile, Bing Places, Apple Business), and keep name, address and phone consistent.
  4. Publish or tidy obvious public signals: relevant schema.org types on the homepage, and /llms.txt if you want one. Google does not require llms.txt or a special schema type for AI Overviews or AI Mode.

The fixes are unusually cheap relative to most digital work. First, get an honest baseline: know which of the seven dimensions are genuinely weak and which are merely invisible. Second, close the invisibility gap where the capability already exists: a short, factual AI or technology statement; a data-handling position beyond the boilerplate privacy policy; job adverts that name the data and AI skills you actually want. Third, fix the genuinely weak foundational dimension (for most SMEs, data or governance) before spending on tools.

Each of those is a one-day task, and each changes what every future buyer, investor and search engine reads about you.

About this data

234 UK SME and mid-market brands scanned with Z3nith's public-signals assessment and stored in our research dataset for longitudinal follow-up. The sample is balanced at up to 22 brands per sector for a fair cross-sector comparison; two sectors fall below that line, education and training (20) and wholesale and logistics (16), simply because the pool of genuinely well-known UK brands there is smaller and a number of business-to-business sites block automated access. Companies are not named: the point is the pattern, not a league table. Scores reflect externally visible evidence only. A verified assessment with the companies' own input would score materially higher (completing the questionnaire typically moves scores by 15 to 30 points). We re-run the study quarterly; sector-level deep-dives are available on request.

June figures on this page are the published June 2026 scan. November figures are the November 2026 public-signals scan of the same 234-brand published cohort. The published cohort is the first 22 successful scans per sector, ordered by scan time and then by domain, except education and training (20) and wholesale and logistics (16). The formula, the weights and the seven-dimension set are unchanged since before the June batch.

Cohort-level June versus November totals, and the sector means, are the comparison this page makes. Brand-level June to November movement is not safe, because of a change on 17 October 2026 to Companies House auto-matching. That change skips .gov.uk, .nhs.uk, .ac.uk and similar domains, and it tightens entity ranking. This page does not name companies, and it does not list who rose or who fell.

The score is what is publicly visible. It does not measure whether an AI product will recommend a business. Allowing a crawler is not a guaranteed AI Overview, and it is not a guaranteed citation in an AI answer.

Frequently asked questions

What is a public-signals AI-readiness score?

A score built only from what is publicly visible: the company website and technology stack, job adverts, Companies House filings and published policies. It measures what buyers, investors and AI search engines can see, and is a conservative baseline rather than a verdict on internal capability. It does not measure whether an AI product will recommend the business.

Why did 234 well-known UK brands all score below 20/100?

Because the dimensions that drive AI readiness (strategy, governance, talent) are almost never communicated publicly. The average visible score was 10.2/100 while the same companies averaged 61/100 on AI opportunity. Most are far more capable than they appear; the gap is invisibility as much as unreadiness. The November 2026 scan of the same cohort also had none at 20. Mean visible readiness was 8.9.

Which UK sector has the biggest AI opportunity gap?

In the June 2026 scan, construction and architecture in this balanced study: average visible readiness of 9.5/100 against an average AI opportunity of 64.5/100, a sector rich in schedule, cost and design data whose public presence barely reflects it. Legal services is close behind at 9.8 against 60.8. That is the June comparison. Every November sector mean is lower than June. November sector means run from 6.9 (food and beverage) to 10.4 (professional services). This page does not name companies.

How can an SME improve its externally visible AI readiness?

Run the free scan. Free. No credit card required. No sign-up. Check crawler access for GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot and Google-Extended, and fix a blanket disallow if one is in place. Claim or update Businesses in ChatGPT / Business Hub, Google Business Profile, Bing Places and Apple Business, and keep name, address and phone consistent. Publish relevant schema.org types on the homepage, and /llms.txt if you want one. Google does not require llms.txt or a special schema type for AI Overviews or AI Mode. Also publish a short factual AI or technology statement, a data-handling position beyond the standard privacy policy, and job adverts naming the data and AI skills you want, then verify your underlying readiness with a structured assessment so the claims are evidenced.

What did the November 2026 scan find?

The November 2026 public-signals scan of the same 234-brand published cohort found mean visible readiness of 8.9, a median of 8.6, and a range of 5.1 to 15.5. The published June 2026 scan was a mean of 10.2, a median of 10.1, and a range of 5.1 to 17. Bands for 5 to 9.9, 10 to 14.9, 15 to 19.9 and 20 or above were 166, 66, 2 and 0 in November, against 116, 112, 6 and 0 in June. None reached 20 in either batch. Sector means are the comparison this page makes. Company-by-company movement is not, because Companies House auto-matching changed on 17 October 2026.

Get your AI readiness score

Paste your company URL. In about 60 seconds you get a unique results page and a headline score out of 100, from what you publish.

Run a free assessment