Quarterly research
SCAN DATA · UPDATED HOURLY
The state of agent readiness
Technical scan results from AgentReady Check, updated hourly. Each domain contributes its latest scan under the current scoring version.
597 domains · 1,000 scans
597 domains in this sample.
Latest scan per domain · Current scoring version
Headline numbers
48/100
Average agent-readiness score
3 %
Name AI crawlers explicitly in robots.txt
5 %
Serve a working llms.txt
26 %
Are near-empty without JavaScript
63 %
Have no structured data at all
36 %
Have every agent basic in place
76 %
Show provable search index verification
12 %
Of the domains that sell something, have usable product data
Score distribution
- 0–39
- 170
- 40–69
- 363
- 70–100
- 64
Average result per area
Each average is a percentage of applicable points. Scans where an area did not apply are excluded from that average.
RAI crawler access
61 %
RJavaScript dependency
72 %
RAgent basics
66 %
RSearch index verification
86 %
Ellms.txt
14 %
EStructured data
19 %
EFeed & product data
23 %
AAgentic buyability
45 %
DCatalog prerequisites
22 %
How these numbers are derived
These figures use the public Check scoring. Each percentage below counts domains that meet a threshold in an applicable area.
- AI crawler access: at least 80% of applicable points, requiring named crawlers in robots.txt and matching status codes for crawler and browser requests.
- Working llms.txt: at least 80%, requiring a 200 response, a text content type and no HTML soft 404.
- Near-empty without JavaScript: below 40% in the JavaScript area, indicating well under 1,500 characters of visible text before scripts run.
- No structured data: zero points for structured data, meaning no usable JSON-LD on the homepage.
- All agent basics in place: full applicable points for sitemap, canonical URL, legal or contact links and response time.
- Search verification: a Google or Bing verification found in DNS or HTML. File uploads and Tag Manager verification are outside the scan, so this percentage is a lower bound.
- Usable product data: at least 60% in feed and product data, calculated only among domains where that area applies.
Each domain is counted once, using its most recent scan, so a domain scanned twenty times cannot bend the result. Scans are anonymous: we store the domain, the score and the check results, never who ran the scan.
In September 2026 the Check expanded from five areas to nine and changed their weights. This report includes only scans using the current scoring version, so its sample can be smaller than the lifetime scan count.
The public scoring is described under How it works. The same data is available as JSON at /api/v1/stats, free and without a key.
Also available as Markdown.