Methodology

How we measure a website's readiness for AI agents - honestly and reproducibly.

What "agentic readiness" means

An AI agent reads a page the way a machine does: through the accessibility tree, structured data and discovery files, not a rendered screenshot. Agentic readiness measures how well a site supports that, plus the trust signals an agent (or a person) weighs before transacting. Our checks mirror Google's Lighthouse "agentic browsing" scoring.

How we score

Like Lighthouse, we do not emit an invented 0-100. Each page is measured against 30 deterministic checks; every check passes, partially passes, or fails based on the site's real HTML and discovery files. The headline is a fraction of passed checks. Where a signal cannot be assessed, we say so.

The checks

Agent interaction

• semantic html
• clean html
• extraction friction
• indexability
• layout stability
• canonical url
• document weight
• critical path efficiency

Machine extraction

• table list extractability
• sentence atomicity
• entity density
• definition patterns
• answer capsule pattern

Structured data

• schema markup
• schema coverage
• speakable schema

AI discovery

• llms txt
• robots txt
• sitemap completeness
• rss feed
• internationalization signals
• content licensing

Trust & transparency

• entity consistency
• creator transparency
• methodology transparency
• ai disclosure
• visible date signal
• title meta quality
• content freshness
• author schema depth

Data & provenance

Checks are computed from a site's own homepage and its discovery files (llms.txt, robots.txt, sitemap.xml). Each report shows a last-checked date, and that date drives the page's freshness signal in our sitemap. Scores reflect the site and can be improved by its owner, never bought.