Quarterly research
A technical check of company websites across Europe, the United States and Japan. What agents can read, where access falls short, and how the results differ by industry.
01 / The sample
Companies selected from public rankings across five industries. The panel stays fixed between editions, with changes documented.
This check looks for named AI crawlers in robots.txt and matching responses to crawler and browser requests. It was assessable on 185 of 185 companies. It does not establish why a company allows or restricts access.
Among 185 assessable companies, this share scored zero on the structured-data check. Product details or contact information may still exist in page text or other formats.
Among 185 assessable companies, this share scored below 40% on the JavaScript check. Clients that read the initial HTML may miss content loaded later by scripts.
Average scores are 56.0 across 38 companies and 35.3 across 38. This compares the selected panel groups; it is not an estimate for the whole industry.
These companies reach the highest score band on the technical check. The score measures applicable public website signals; it does not test whether an AI assistant recommends the company or completes a purchase.
Score distribution
The four pillars
R · Retrievable
66 %185 of 185
E · Explainable
20 %185 of 185
A · Actionable
51 %13 of 185
D · Distributed
20 %7 of 185
Average share of applicable points. Each row shows how many companies could be assessed. A missing result is excluded from that average.
Regional comparison
Average scores on the same 0–100 scale. Each region contains a different mix of companies.
92 companies · median 53 · middle half 26–65
93 companies · median 56 · middle half 29–66
Industry comparison
| Segment | Companies | Average | Median | Middle 50% |
|---|---|---|---|---|
| Finance & insurance | 38 | 56.0 | 61 | 48 – 68 |
| Local service chains | 37 | 55.6 | 57 | 48 – 66 |
| SaaS | 36 | 55.6 | 62 | 43 – 67 |
| D2C & e-commerce | 36 | 44.0 | 44 | 26 – 63 |
| Travel | 38 | 35.3 | 30 | 23 – 44 |
AgentReady Index
Only the published top performers are named. The rest of the panel appears in aggregate.
Europe · Top 10
Scores out of 100, using the same scoring version as the rest of this panel.
United States · Top 10
Scores out of 100, using the same scoring version as the rest of this panel.
The panel is drawn from public rankings and fixed for each edition; additions and removals are documented per edition. You can measure your own site with the same check in about ten seconds.
Percentages below are average shares of applicable points. Coverage varies by check: some checks do not apply or could not be assessed.
65%
Average points · 185 of 185 assessed61%
Average points · 185 of 185 assessed22%
Average points · 185 of 185 assessed59%
Average points · 185 of 185 assessed20%
Average points · 185 of 185 assessed92%
Average points · 185 of 185 assessed26%
Average points · 13 of 185 assessed51%
Average points · 13 of 185 assessed20%
Average points · 7 of 185 assessed02 / The sample
Every member of the S&P 100, the EURO STOXX 50 and the Nikkei 225. Membership is the only criterion, so there is no selection to defend.
This check looks for named AI crawlers in robots.txt and matching responses to crawler and browser requests. It was assessable on 373 of 373 companies. It does not establish why a company allows or restricts access.
Among 373 assessable companies, this share scored zero on the structured-data check. Product details or contact information may still exist in page text or other formats.
These companies reach the highest score band on the technical check. The score measures applicable public website signals; it does not test whether an AI assistant recommends the company or completes a purchase.
Score distribution
The four pillars
R · Retrievable
68 %373 of 373
E · Explainable
10 %373 of 373
A · Actionable
38 %3 of 373
D · Distributed
20 %2 of 373
Average share of applicable points. Each row shows how many companies could be assessed. A missing result is excluded from that average.
Regional comparison
Average scores on the same 0–100 scale. Each region contains a different mix of companies.
50 companies · median 54 · middle half 39–65
100 companies · median 52 · middle half 34–62
223 companies · median 47 · middle half 41–53
Industry comparison
| Segment | Companies | Average | Median | Middle 50% |
|---|---|---|---|---|
| Health care | 32 | 50.1 | 53 | 39 – 62 |
| Financials | 55 | 50.0 | 52 | 42 – 62 |
| Technology | 83 | 47.9 | 48 | 37 – 61 |
| Energy & utilities | 19 | 46.5 | 49 | 43 – 59 |
| Industrials | 115 | 46.4 | 47 | 42 – 56 |
| Consumer | 69 | 45.4 | 48 | 36 – 54 |
AgentReady Index
Only the published top performers are named. The rest of the panel appears in aggregate.
Europe · Top 10
Scores out of 100, using the same scoring version as the rest of this panel.
United States · Top 10
Scores out of 100, using the same scoring version as the rest of this panel.
Japan · Top 10
Scores out of 100, using the same scoring version as the rest of this panel.
The panel is drawn from public rankings and fixed for each edition; additions and removals are documented per edition. You can measure your own site with the same check in about ten seconds.
Percentages below are average shares of applicable points. Coverage varies by check: some checks do not apply or could not be assessed.
74%
Average points · 373 of 373 assessed65%
Average points · 373 of 373 assessed15%
Average points · 373 of 373 assessed61%
Average points · 373 of 373 assessed6%
Average points · 373 of 373 assessed83%
Average points · 373 of 373 assessed26%
Average points · 3 of 373 assessed38%
Average points · 3 of 373 assessed20%
Average points · 2 of 373 assessedAbout the research
A fixed panel, public website signals, and a scoring version recorded with every edition.
Every company is measured with AgentReady Check. The checks cover crawler access, HTML before JavaScript, agent basics, search verification, llms.txt, structured data, feeds, actions and distribution. The report uses technical scan results only.
558 of 558 panel entries produced a result. A domain may appear in both panels. These are selected groups, so their results do not represent all websites in Europe, the United States or Japan.
This edition uses scoring version 2. Comparisons require the same scoring version. If the method changes, the previous panel must be rescored before a trend can be reported.
Up to 10 top performers are named where a group contains at least 20 companies. Other individual scores stay private. Segments below 5 companies are omitted; small distribution groups are merged to contain at least 3.
Read the full report as Markdown · View the hourly scan sample
Your website
Get your technical score and see which checks need attention.