Scoring Methodology
The Readiness Score is /100, built from four evidence-weighted pillars. Every check below links to the published research behind its weight — and this page is generated from the same registry the auditor runs, so it can never drift from what we actually measure.
The four pillars
Raw check points inside each pillar are normalized to the pillar's weight, so adding or re-weighting checks never changes the meaning of the headline score. Business-type multipliers (SaaS, e-commerce, media, SMB) adjust individual checks before normalization. Tiers: 80+ AI-Ready, 60–79 AI-Aware, 40–59 AI-Exposed, below 40 AI-Blind.
Access & Retrievability — 30 of 100
| Check | Weight | Evidence |
|---|---|---|
AI retrieval bots can fetch pages (CDN/WAF level) | 8 pts | Cloudflare pay-per-crawl announcement |
Content visible in raw HTML (no JavaScript required) | 6 pts | Vercel/MERJ AI-crawler rendering study |
No restrictive snippet/preview controls | 4 pts | Shepard 54-study meta-analysis |
XML sitemap present and valid | 3 pts | hygiene / supporting |
Primary CTA accessible without CAPTCHA Weighted by business type: smb ×0.6, ecommerce ×1, saas ×1, media ×0.7 | 3 pts | hygiene / supporting |
robots.txt present and valid | 2 pts | hygiene / supporting |
Canonical tags present | 2 pts | hygiene / supporting |
Contact/booking form accessible without hard bot block Weighted by business type: smb ×1, ecommerce ×1, saas ×0.8, media ×0.6 | 2 pts | hygiene / supporting |
OpenAI bot stance documented in robots.txt | 1 pts | BuzzStream training-vs-retrieval blocking study |
Anthropic bot stance documented in robots.txt | 1 pts | hygiene / supporting |
Content & Extractability — 35 of 100
| Check | Weight | Evidence |
|---|---|---|
Answer-first structure (content leads the page) | 6 pts | Shepard 54-study meta-analysis |
Statistics and quotations density | 6 pts | Princeton GEO paper (KDD 2024) |
Content chunked into heading-led sections | 4 pts | Shepard 54-study meta-analysis |
Answer-dense content patterns present | 4 pts | Shepard 54-study meta-analysis |
Content freshness (visible, recent dates) | 4 pts | Shepard 54-study meta-analysis |
Specific verifiable claims present | 3 pts | Princeton GEO paper (KDD 2024) |
Named author attribution present | 3 pts | hygiene / supporting |
Sufficient unique content (low duplication risk) | 3 pts | hygiene / supporting |
Schema.org structured data present | 2 pts | Ahrefs schema causal study |
FAQPage or HowTo schema present | 2 pts | hygiene / supporting |
OpenGraph tags complete (title, description, image) | 2 pts | hygiene / supporting |
Attribution & Measurement — 10 of 100
| Check | Weight | Evidence |
|---|---|---|
Analytics platform installed and active | 4 pts | hygiene / supporting |
Noma Intel Smart Tag installed | 4 pts | hygiene / supporting |
UTM parameters present on key links | 3 pts | hygiene / supporting |
Reported, deliberately not scored
These observations appear in every report as “AI Posture & Compliance”, but carry zero score weight — because the evidence says they don't move AI visibility. llms.txt: no major engine reads it (97% of published files receive zero bot requests). Training-bot opt-outs: blocking GPTBot or Google-Extended does not remove you from AI answers — it's a data-governance choice, not a visibility factor, and rewarding it in a visibility score would be contradictory. Emerging agent standards (agent.json, MCP): promising, unmeasured — we'll re-weight when evidence lands. Privacy posture: a compliance matter, reported for completeness.
Show all 19 informational checks
/llms.txt present | — | Ahrefs llms.txt study |
/llms-full.txt present | — | Ahrefs llms.txt study |
Google-Extended stance documented | — | hygiene / supporting |
CCBot (Common Crawl) stance documented | — | hygiene / supporting |
Bytespider, PerplexityBot, YouBot covered in robots.txt | — | hygiene / supporting |
GPTBot blocked (OpenAI training opt-out) | — | BuzzStream training-vs-retrieval blocking study |
Google-Extended blocked (Gemini training opt-out) | — | BuzzStream training-vs-retrieval blocking study |
CCBot blocked (Common Crawl training opt-out) | — | hygiene / supporting |
Terms of Service references AI / automated scraping | — | hygiene / supporting |
Privacy Policy references AI data processing | — | hygiene / supporting |
/.well-known/agent.json manifest present Weighted by business type: smb ×0.4, ecommerce ×0.6, saas ×1, media ×0.5 | — | hygiene / supporting |
MCP (Model Context Protocol) server published Weighted by business type: smb ×0.3, ecommerce ×0.5, saas ×1, media ×0.4 | — | hygiene / supporting |
OpenAPI / Swagger spec published Weighted by business type: smb ×0.2, ecommerce ×0.6, saas ×1, media ×0.3 | — | hygiene / supporting |
Cookie consent banner present | — | hygiene / supporting |
On-site AI features disclosed | — | hygiene / supporting |
Privacy Policy references AI / automated processing | — | hygiene / supporting |
Third-party AI service data sharing disclosed | — | hygiene / supporting |
Data retention period stated in Privacy Policy | — | hygiene / supporting |
Contact email listed for data/privacy queries | — | hygiene / supporting |
What this score is not
The Readiness Score measures what you control on and around your site — it is not a measurement of how often AI engines actually mention you. Engine answers are volatile (the same query returns ~9% overlapping results across runs in some engines), so a defensible visibility number needs repeated sampling with confidence ranges. That measured Visibility Score is coming as a separate number beside this one — we won't blend them, because hiding the cause-and-effect between the two is how tools mislead.
Weights are reviewed quarterly against new research. When the evidence changes, the score changes, and this page changes with it — automatically.