How the score is calculated

The AI-Ready Audit assigns 0 to 100 points from five automatically measured criteria. This page exists so the number can be audited, disputed and cited — a score whose arithmetic nobody can check is not a reference.

The five criteria and their weights

• Performance

  Weight 40

  The Lighthouse performance score (0–1) is converted proportionally into the 40 points. Above 70 counts as good; 40 to 70, warning; below 40, critical.

  Data source: Google PageSpeed Insights (mobile strategy)

• llms.txt

  Weight 20

  Binary: the file exists and responds, or it does not. We do not judge its content — judging text quality would require an opinion, and this criterion has to be reproducible by anyone.

  Data source: HTTP request to /llms.txt at the domain root

• Schema markup

  Weight 20

  JSON-LD present earns the full 20. Microdata alone earns 10, being a weaker and fading signal. Neither, zero.

  Data source: Home page HTML

• Essential metadata

  Weight 10

  The meta description must exist and run between 80 and 160 characters. Outside that range it earns half: present, but too short to inform or too long to display in full.

  Data source: Home page HTML

• HTTPS

  Weight 10

  Binary. A prerequisite rather than a differentiator — hence the low weight — but its absence is flagged to the visitor before they read anything.

  Data source: Scheme of the final URL, after redirects

Reading bands

80–100

Ready for the agent era

50–79

In development

0–49

Action needed

Known limits of the method

• Only the home page is analysed. A site can have exemplary schema on inner pages and still score poorly here.
• When a criterion cannot be measured — the PageSpeed API being unavailable, for instance — it is EXCLUDED from the denominator instead of counting as zero. The result says it is partial and should not be compared with a complete run.
• The weights are an editorial judgement by Infinity, not an industry standard. Performance carries 40 because it is the only criterion with measured impact on both ranking and conversion.
• The llms.txt criterion measures adoption of a still-young convention. Its absence does not make a site invisible to AI — but its presence removes ambiguity about what the company does.
• The analysis reads only what is already public. Nothing is executed on the evaluated site and no credentials are requested.

Versioning

This is version 1.0, in force since 2026-07-31. Changes to criteria or weights raise the version, because changing them silently would invalidate every comparison already published.