AEO Audit
The 330-check AEO technical audit
This is the register I run a site against to find what is suppressing its rankings and AI citations. Every check is clustered into the four pillars, cross-referenced to a primary source, and written so you can see what it is, why it matters, how to fix it and how to find it yourself.
AI search does not run on a separate index or a special file you feed it. It runs on the same crawl, render and index pipeline as classic search, layered with stronger trust and entity signals. So I audit across four pillars: can a machine reach the page, does the page answer the query, is there a real expert behind it, and can a machine connect it to a known entity. These 330 checks are how I make that concrete.
The four pillars
Technical Foundations
The first pillar decides whether Google and AI crawlers can reach, render and index a page in the first place. Robots directives, status codes, canonicalisation, JavaScript rendering, Core Web Vitals, HTTPS, sitemaps and URL structure. If this pillar fails, nothing downstream matters, because the page is never eligible to be ranked or quoted.
Topical Content
The second pillar is about whether each page answers a real query well enough to be cited. Titles and meta, heading structure, body depth and originality, internal linking, images and rich media. This is where a page earns the right to be the answer rather than just another result.
Trust Signals
The third pillar is the evidence of experience, expertise, authoritativeness and trust that tells Google and AI models a real, accountable person or organisation stands behind the page. Authorship, bylines, credentials, about and contact pages, and site-level trust signals.
Authority Network
The fourth pillar is the machine-readable layer: structured data validity, specific schema types, entity and Knowledge Graph signals, local and ecommerce markup, off-site authority and agentic readiness. This is what lets a machine understand what a page is about and connect it back to a known entity.
Guardrails and validity
The cross-cutting pillar covers hygiene and future-proofing that does not belong to any single pillar: valid HTML, XML and JavaScript, avoidance of manipulation and spam patterns, and resilience as search continues to shift towards AI answers.
How to read a check
Every check carries a severity and a confidence tag, because I would rather tell you how sure I am than pretend everything is settled fact. Verified means Google states it directly. Inferred means I have reasoned it from a Google source. Tactical means it is my own recommendation from doing the work. Industry consensus means it is widely accepted practice. Where no primary source exists, the check says so rather than inventing one.
The field notes
Frameworks, builds and what is changing in AI search.
Sent occasionally, never noise. The thinking behind the work, and the experiments before they ship.
No spam. Unsubscribe anytime.