Access: checks 1–3
1. Check the intended search crawler for the exact URL. OAI-SearchBot concerns ChatGPT Search; GPTBot concerns training. Google Search uses Googlebot. Google-Extended controls separate Gemini uses and does not establish Search eligibility.
2. Compare the initial HTML with the rendered page. A static scanner cannot assess elements that are only created later. Do not assume all providers process JavaScript identically.
3. For Google, verify indexing, snippet eligibility and the Search Console generative AI inclusion setting. Public HTML alone cannot confirm these conditions.
Useful content and structure: checks 4–6
4. Publish evidence only you can provide: a documented test, original data, an experienced explanation or a reproducible case study.
5. Use headings and navigation that help readers find the right passage. There is no required 200-word chunk size or ideal AI keyword quota.
6. Describe visible entities accurately if you use structured data. Google requires no special AI schema or llms.txt file for its Search AI features.
Attribution and verification: checks 7–10
7. Identify the author or maintaining organization accurately and link to the relevant background.
8. Link factual claims to sources that actually support them. Separate controlled experiments from customer outcomes and broader statistical research.
9. Keep publication and revision dates honest. Update a date when the content changes substantively.
10. Record actual citations, Search Console impressions, qualified referrals and completed actions. These answer a different question from a technical diagnostic score.
Read a measured example, then check your page
The four-audit study shows how the same guide changes under noindex, missing accessible names and script-only content. All source and checks are in the sample report.
Run the AI readiness checker to inspect a public URL. The overall score is visible without signup; email unlocks the full report. The daily allowance is three successful analyses per network.