Open methodology

AI Visibility Score — Methodology V3.4

The AI Visibility Score measures how well generative AI systems — ChatGPT, Claude, Perplexity, Gemini — can reach, understand, verify and cite your website. Nine building blocks produce a raw score of roughly 137 points, normalised and capped onto a 100-point scale.

Technical AI accessibility carries the most weight — if AI bots can't reach or parse your content, nothing else matters. For the same reason, penalties apply when fundamental signals are actively broken.

A from 80B from 60C from 40D from 25F below

1. Foundation

40 points

Can AI bots reach and understand your site?

  • Organization schema — Base identity: who are you? The highest-weighted single check.10 pts
  • FAQPage schema — Strongest correlation with AI citations.8 pts
  • AI-friendly robots.txt — Present, and GPTBot, ClaudeBot, PerplexityBot, Google-Extended not blocked.5 pts
  • BreadcrumbList schema — Navigation hierarchy for AI.5 pts
  • sameAs verification — At least 3 links to Wikipedia, Wikidata, LinkedIn, X = 5 pts; 1–2 links = 2 pts.up to 5 pts
  • Speakable schema — Readability for voice assistants.3 pts
  • Canonical tag2 pts
  • Hreflang2 pts

2. Sector schemas

25 points

Do you speak your industry's language?

  • Required schemas — 5 pts per required schema for your sector. E-commerce example: Product + Review.up to 10 pts
  • Valuable schemas — Pro rata by completion, e.g. GTIN, shipping, returns policy.up to 10 pts
  • Bonus schemas — 2.5 pts per sector-specific extra.up to 5 pts

3. Reviews & trust

10 points

Can AI verify your credibility?

  • AggregateRating / Review schema — Machine-readable rating signals as a trust anchor.10 pts

4. Forward-thinking

21 points

Are you ready for next-gen AI standards?

  • llms.txt — AI-readable sitemap. Only the top 1% of sites have one.4 pts
  • llms-full.txt — Extended AI content file.3 pts
  • AI Transparency Protocol — Machine-readable AI transparency.3 pts
  • HowTo schema3 pts
  • UCP manifest — Unified Commerce Protocol, live since Jan 2026 at Google, Shopify, Target.2 pts
  • WebMCP — AI agents use site functions directly. W3C draft 2026.2 pts
  • ai.txt — Emerging AI permission standard.1 pt
  • IndexNow — Bing/Yandex instant indexing.1 pt
  • AI content quality — Low hallucination and spam risk = 2 pts, medium = 1 pt.up to 2 pts

5. Agentic protocols

6 points

Can AI agents transact with your site?

  • ACP protocol — Agentic Commerce Standard by Stripe and OpenAI. Scored for commerce sectors only.3 pts
  • MCP server card — /.well-known/mcp.json2 pts
  • Agent skills — /.well-known/agent-skills.json1 pt

6. ARP protocol

15 points

Do you control how AI reasons about your brand?

  • reasoning.json — Sub-score of the Agentic Reasoning Protocol. Teaches AI your brand logic, sets anti-hallucination guardrails and recommendation boundaries. Engine-independent and the most durable lever in the whole model.up to 15 pts

7. Schema diversity

5 points (bonus)

How rich is your structured markup?

  • Valid JSON-LD schemas — Maximum 5 pts. Only error-free, parsable schemas count.2 pts each

8. Crawler access health

8 points

Is your robots.txt cleanly configured for AI crawlers?

  • Sub-score — Evaluates AI-bot policy, sitemap directive, syntax errors and blanket disallow rules.up to 8 pts

9. Performance

up to 8 points

Does Googlebot truncate your page before it understands it?

  • HTML size within budget — Active. Googlebot truncates above 2 MiB.2 pts
  • LCP / INP / CLS / TTFB — Arriving with the CrUX and Lighthouse integration in V3.5. Currently only HTML size contributes.pending

Penalties

  • Cloudflare blocks AI bots — WAF doesn't let GPTBot, ClaudeBot and others through.−3
  • High AI content risk — Content with high hallucination or spam risk.−5
  • Invalid JSON-LD schemas — Invalid JSON is ignored by crawlers.−3

Sector detection is automatic: we identify your sector — e-commerce and its specialisations, local service providers, creative and photography, generic fallback profile — from domain patterns and detected schema types, then apply sector-specific requirements.

Methodology version: Scoring V3.4 · Sector profiles V3.0

See your own score.

The same methodology applied to your domain — in under 60 seconds, no signup.

Free scanDiscuss a project