1. Fetch, render, and URL integrity
The gate. If these fail, nothing below matters, which is why this section carries the heaviest weight in the score.
2. Bot access and control plane
Whether you have invited engines in or locked them out, sometimes by accident.
3. Off-site presence and mentions
What the rest of the web says. The hardest section to fake and the slowest to fix, which is why it starts now.
4. Structured data
The weakest dimension in the benchmark: the average audited site scores 49/100 here and a quarter score 5 or below, effectively shipping nothing.
5. HTML extractability and main content clarity
Whether a parser can pull clean text without fighting your markup.
6. Entity clarity
Whether a machine can tell who you are. Ambiguity here is how engines recommend your competitor while describing your product.
7. Trust and security
Accountability signals plus the security surface. Engines prefer recommending sites that look operated.
8. Content freshness and authority
The section where the benchmark's top decile separates hardest from the median: 32 points of gap.
9. Content answerability
Whether the writing is shaped so an engine can lift an answer.
Most sites do not lose AI visibility on the hard problems; they lose it on un-shipped basics that take an afternoon each.
How to work the list
Sequence by section, not by ease
Most items are afternoons
Re-check quarterly
What to do next
Run the free audit and let it mark this list for you, with your actual pass-fail per check and the fixes ranked by weight.