aeo-page-audit

AEO Page Audit

Answer engines don't rank pages, they extract answers and cite sources. This audit checks whether a page is extractable, and whether it deserves to be the citation.

Step 0: Define the prompts to own

Before auditing anything, ask: which 3–5 buyer prompts should this page win? ("best <category> for <segment>", "how do I <job>", "<product> vs <alternative>", "<category> pricing"). Write them down. Every finding is judged against these prompts, an AEO audit without target prompts is just an SEO audit with new jargon.

If possible, test the prompts live in ChatGPT/Perplexity and record: is the user's site mentioned, who is cited instead, and what those cited pages have that this one lacks.

The audit, four layers

Layer 1: Access, can bots read it at all?

• Fetch the page. Does meaningful content arrive in the raw HTML, or is it client-rendered into an empty shell? (AI crawlers execute little to no JS.)
• Check robots.txt for blocks on GPTBot, OAI-SearchBot, PerplexityBot, ClaudeBot, Google-Extended, blocking them is opting out of citations.
• Gated/paywalled content: invisible. Gate the asset, never the explanation.
• Bonus signals: llms.txt present; sitemap current; clean canonical.

Layer 2: Extractability, can a model lift the answer?

• Answer-first structure: does each section open with the answer in the first sentence, then support it? Flag sections that build suspense.
• Question-shaped headings: H2/H3s that match how buyers phrase prompts.
• Self-contained blocks: could a paragraph be quoted alone and still make sense? Flag pronouns pointing at prior sections ("this approach", "it").
• Facts with numbers: models prefer citing specifics, prices, timeframes, counts, dates. "Fast onboarding" is uncitable; "onboarding in 9 days" isn't.
• Structured data: FAQPage/Product/Service/Organization JSON-LD present and matching visible content.

Layer 3: Answer quality, does it deserve the citation?

For each target prompt: does the page actually contain the best available answer, stated more clearly and specifically than the currently cited competitors? Generic content loses to specialized content in AI answers even harder than in classic search. Flag sections that say what every competitor says.

Layer 4: Surround sound, is the entity known off-site?

Models triangulate. Check (and list gaps): consistent company description across site/LinkedIn/directories, presence in relevant "best X" lists and comparison pages, review sites for the category, community mentions (Reddit, niche Slacks, HN where relevant). One page can't do AEO alone.

Output format

## AEO Audit: <URL>
**Target prompts**: <the 3–5 prompts>
**Live test**: <cited today? who wins instead, and why, one line each>

### Fix now (blocking)
### Fix next (extractability)
### Content gaps (answer quality)
### Off-site gaps (surround sound)

Each finding: what's wrong → evidence from the page → the specific change, with rewritten copy where relevant. End with the one change most likely to earn a citation within 90 days.

Related skills

positioning-canvas for the claim worth being cited for. founder-content-engine to build the evidence. landing-page-teardown once the traffic arrives.