AI Readiness Score methodology

Model v1. A single, stable, explainable 0–100 grade for how well a site works for AI crawlers (GPTBot, ClaudeBot, Perplexity, Gemini, Grok, Doubao). We publish the model so scores are reproducible and citable.

Category weights

CategoryWeightWhat it measures
Crawlability0.35Can AI crawlers fetch you at all? Highest weight — being blocked is fatal.
Renderability0.18Is your content in the raw HTML, or hidden behind JavaScript?
Structured data0.17JSON-LD / schema.org that lets AI parse entities.
Content accessibility0.15llms.txt, canonical, and HTTP headers.
Performance0.15Response time and payload size.

How it's computed

Each category starts at 100 and loses points per issue found:

  • Critical issue: −30
  • Warning: −10
  • Info: −3

The overall score is the weighted sum of the five category scores. Then gating caps apply — because being blocked is fatal for AI visibility:

  • Any critical Crawlability issue (e.g. robots.txt or Cloudflare blocking a major AI crawler) caps the overall score at 45 (F).
  • Any critical Renderability issue (content invisible without JavaScript) caps the overall score at 72 (C).

Grade bands

A+97–100
A90–96
B80–89
C68–79
D50–67
F0–49

Bands are strict on purpose — most unoptimized sites land B or C. Grades are shown with letter + colour + text for accessibility.

Per-agent scores

Each of the six AI crawlers gets its own grade based on the issues that affect it: −30 for a high-impact issue, −12 medium, −4 low. This is why a site can be great for one crawler and blocked for another.

Versioning

Every score records the model version it was computed with (currently v1). When we change weights or bands we bump the version and announce it, so historical scores and leaderboards stay comparable.