EU AI Readiness Observatory · Report 01
A high SEO score does not predict AI readiness
Strong SEO does not guarantee that AI systems can use a company website as a source. Across 78,110 EU-27 websites, 68.6% of strong-SEO sites remain below the citation-readiness threshold.
Correction note (31 August 2026): version 1.1 overstated L1=0 by 8,021 websites (29.1% instead of 18.8%); version 1.2 recomputes every figure, while the core conclusion—strong SEO does not close the citation-readiness gap—remains unchanged.
What this means for leaders
Your SEO dashboard does not show whether AI systems can use your company website as a source. In this dataset, 68.6% of strong-SEO websites remain below the citation-readiness threshold, and the gap reaches 24.4 points in the 90+ SEO band.
If competitors publish clearer, easier-to-use facts and evidence, AI systems may find them more useful as sources.
Ask three questions
- Can crawlers obtain usable content?
- Can the site clearly state what the company does, for whom and where?
- Are important claims backed by current, extractable evidence?
What we measured
Every company website in the Observatory is assessed across eight layers. The EU cohort is defined by headquarters country in firmographic data, never by domain extension.
All 78,110 websites stay in the market sample. L1 records whether the crawler could obtain usable page content in the measured visit. When L1=0, that inability to access the site is itself the result. L2–L8 averages are then calculated only for L1>0 websites because there is no downstream page content to analyse at L1=0.
| Sample | Websites | How it is used |
|---|---|---|
| Full EU-27 firmographic cohort | 78,110 | Crawler-access market result |
| Crawler access L1 = 0 | 14,708 (18.8%) | Retained in the full sample as not crawler-accessible |
| Crawler access L1 > 0 | 63,402 | Content-layer and cross-layer analysis |
Why L1 was zero
| L1=0 diagnostic reason | Websites |
|---|---|
| Server returned HTTP 5xx | 3,580 |
| WAF or firewall prevented access | 3,457 |
| No usable server-rendered content was returned | 3,347 |
| The site declared noindex | 1,214 |
| HTTP access failed without a more specific diagnosis | 852 |
| Authentication was required | 690 |
| More than one access failure was detected | 663 |
| robots.txt denied crawler access | 587 |
| CDN challenge or redirect prevented access | 170 |
| The server rate-limited the crawler | 73 |
| No readable page structure was available | 69 |
| No diagnostic cause was recorded | 6 |
The diagnostic reasons sum to 14,708. They explain what the crawler observed; they do not remove those websites from the full market sample.
Finding 1 — SEO explains about a third of AI readiness
Across 63,402 crawler-accessible websites, the correlation between the SEO composite and AI composite is r = 0.567. Between SEO and citation readiness alone it is r = 0.480.
SEO accounts for approximately 32% of the variation in AI readiness. The relationship is positive, but most variation sits outside the classic SEO layers.
Finding 2 — the Citation Gap widens as SEO improves
In the 20–29 SEO band, citation readiness leads SEO by 13.0 points. In the 90+ band, citation readiness trails SEO by 24.4 points.
| SEO band | Websites | SEO | Citation readiness | Citation Gap | AI composite |
|---|---|---|---|---|---|
| 20–29 | 715 | 25.4 | 38.4 | -13.0 | 62.6 |
| 30–39 | 3,020 | 35.2 | 41.6 | -6.5 | 66.0 |
| 40–49 | 7,585 | 45.0 | 45.9 | -0.9 | 69.9 |
| 50–59 | 11,978 | 54.8 | 51.6 | 3.1 | 74.2 |
| 60–69 | 15,964 | 64.8 | 57.2 | 7.6 | 78.0 |
| 70–79 | 16,680 | 74.3 | 61.6 | 12.6 | 81.4 |
| 80–89 | 6,775 | 83.1 | 64.5 | 18.6 | 83.7 |
| 90+ | 605 | 91.9 | 67.6 | 24.4 | 85.9 |
Finding 3 — strong SEO still leaves a citation gap
There are 24,061 websites with an SEO composite of at least 70. Of those, 16,514 — 68.6% — remain below the citation-readiness threshold.
| Segment | Websites | Share | SEO | Citation readiness |
|---|---|---|---|---|
| weak on both | 34,633 | 54.6% | 53.9 | 48.4 |
| strong seo below threshold | 16,514 | 26.0% | 76.8 | 55.9 |
| strong seo above threshold | 7,547 | 11.9% | 78.0 | 77.2 |
| weak seo above threshold | 4,708 | 7.4% | 61.5 | 76.2 |
Finding 4 — the weak group is not technically weaker
Among strong-SEO websites, the group below the citation threshold scores higher on technical SEO: 87.5 versus 83.1. It scores lower on server-side rendering and ground-truth content.
| Strong-SEO group | Websites | SSR L2 | Content L4 | Technical SEO L5 | On-page SEO L6 | Activity L8 |
|---|---|---|---|---|---|---|
| Citation readiness ≥70 | 7,547 | 94.4 | 78.7 | 83.1 | 88.5 | 62.5 |
| Citation readiness <70 | 16,514 | 85.0 | 64.8 | 87.5 | 82.3 | 60.7 |
Finding 5 — the gap exists across European markets
| Country | Websites | SEO | Citation readiness | Citation Gap |
|---|---|---|---|---|
| Czechia | 14,925 | 62.0 | 51.7 | 10.3 |
| Hungary | 511 | 61.4 | 53.1 | 8.3 |
| Portugal | 877 | 65.2 | 57.1 | 8.1 |
| Slovakia | 5,892 | 61.5 | 53.9 | 7.6 |
| Lithuania | 422 | 65.3 | 57.7 | 7.6 |
| Austria | 2,501 | 63.9 | 56.4 | 7.4 |
| Belgium | 888 | 65.7 | 58.4 | 7.3 |
| Sweden | 555 | 66.8 | 59.8 | 7.0 |
| Poland | 1,448 | 64.4 | 57.5 | 6.9 |
| The Netherlands | 2,101 | 66.8 | 60.2 | 6.5 |
| Germany | 15,658 | 61.8 | 55.3 | 6.5 |
| Spain | 4,839 | 66.1 | 59.8 | 6.3 |
| Italy | 4,353 | 64.8 | 59.1 | 5.7 |
| Romania | 721 | 64.5 | 59.1 | 5.5 |
| France | 5,132 | 65.7 | 60.5 | 5.2 |
Czechia has the widest reported Citation Gap at 10.3 points. France has the narrowest at 5.2. Countries with fewer than 400 crawler-accessible websites are omitted from this comparison, not from the full sample.
Finding 6 — where the European market is weak
| Layer | Mean score |
|---|---|
| L8 publishing activity | 45.6 |
| L7 citation readiness | 55.8 |
| L4 ground-truth content | 62.0 |
| L6 on-page SEO | 68.1 |
| L5 technical SEO | 76.2 |
| L2 server-side rendering | 81.6 |
| L1 crawler access | 92.4 |
| L3 indexability | 93.7 |
Publishing activity is the weakest mean layer at 45.6. Indexability is the strongest at 93.7.
What this report does and does not show
- Citation readiness is a structural measurement, not evidence of a live citation in ChatGPT, Gemini, Claude or Perplexity.
- The sample is the VerisAI index filtered to companies with EU-27 headquarters in firmographic data. It is not a random sample of European business.
- All 78,110 company websites remain in the full sample. The 14,708 with L1=0 are reported as not crawler-accessible in the measured visit.
- L2–L8 averages and cross-layer relationships use the 63,402 websites with L1>0 because no downstream page content was available at L1=0.
- Country comparisons require at least 400 crawler-accessible websites; reported SEO bands require at least 100.
- The latest available scan per company at the frozen cut-off is used.
- Correlation is not causation.
Three checks for your company
- Access: can independent crawlers obtain usable content?
- Clarity: does the site clearly state what the company does, for whom and where?
- Evidence: are important claims current, extractable and attributable?
The free AI Readiness Check runs all eight layers on any URL.
Use this data
Aggregated results are published under CC BY 4.0. Reuse is permitted with attribution.
Attribution
VerisAI, EU AI Readiness Observatory, verisai.eu
Cite as
VerisAI (2026). EU AI Readiness Observatory, Report 01: A high SEO score does not predict AI readiness. Corrected edition 1.2, 78,110 EU-27 company websites measured to 31 August 2026. https://verisai.eu/observatory/report-01-seo-does-not-predict-ai-readiness
Download
- report-01.json — Complete metadata, findings and tables
- report-01-by-country.csv — 15 countries
- report-01-by-seo-band.csv — 8 reported SEO bands
- report-01-by-layer.csv — 8 layers
- report-01-segments.csv — 4 segments
- report-01-strong-seo-split.csv — Strong-SEO comparison
- report-01-exclusions.csv — 12 diagnostic reasons for L1=0
- report-01-queries.sql — Frozen cohort and SQL
Per-company data is not published. The licence applies to aggregated results, not the article text.
Frequently Asked Questions
Does this mean SEO no longer matters?
No. The relationship is positive. SEO explains about 32% of variation in AI readiness, but it does not close the Citation Gap by itself.
Are the 14,708 L1=0 websites removed from the report?
No. They remain in the full 78,110-website sample and form the crawler-access result. Only analyses requiring downstream page content use the 63,402 websites with L1 above zero.
Is citation readiness the same as being cited by ChatGPT?
No. It measures whether a site is structurally usable as a source. It does not record live citations.
Which country has the widest Citation Gap?
Czechia at 10.3 points among countries with at least 400 crawler-accessible websites. France has the narrowest reported gap at 5.2.
How is EU defined?
By company headquarters country in firmographic data, never by domain extension.
How large is the sample?
78,110 EU-27 company websites: 14,708 with L1=0 and 63,402 with L1 above zero.
When was the data collected?
The latest scan per company at the frozen cut-off of 31 August 2026.
Run the same eight-layer assessment used in this report on your own site.