How to Audit Your Site for AEO: A 5-Section Audit Framework

Summary — A useful AEO audit covers five sections: crawler access, structured data, content quality, citation graph, and measurement. Done end-to-end, it takes one afternoon and produces a prioritized list of fixes. Done halfway, it produces a false sense of security. This is the framework I use.

For broader context, see What is Answer Engine Optimization?.

What an AEO audit actually answers

A good AEO audit answers four questions in order:

1. Can AI engines reach my content? (Crawler access)
2. Can they parse it cleanly? (Structured data + semantic HTML)
3. Is the content worth quoting? (Quality + format)
4. Does the wider web confirm I'm trustworthy? (Citation graph)

Skip any one and you're auditing in the dark.

Section 1 — Crawler access (15 minutes)

Open yoursite.com/robots.txt. Confirm each of these is not disallowed:

• GPTBot
• OAI-SearchBot
• ChatGPT-User
• PerplexityBot
• Perplexity-User
• ClaudeBot, anthropic-ai, Claude-Web
• Google-Extended
• Googlebot, Bingbot
• Applebot-Extended

For a copy-pasteable reference, see Robots.txt for AI crawlers.

Also confirm:

• Sitemap submitted in Google Search Console and Bing Webmaster Tools
• No accidental noindex on important pages (use a site-wide crawler like Screaming Frog)
• Server returns 200, not soft-404, for canonical URLs

Score: pass / fail. This section gates everything else.

Section 2 — Structured data (30 minutes)

Visit your homepage, product page, and 3 random blog posts. Use a JSON-LD validator (Schema.org's tool, Google's Rich Results Test, or a browser extension) to confirm:

Common audit findings:

• Homepage has no Organization schema at all
• Product page has Offer as a single object instead of an array (loses tier info)
• Blog posts have Article but no dateModified
• FAQ blocks are visible HTML but no FAQPage JSON-LD

Schema priorities are a discipline of their own; the audit's job is to confirm presence and validity, not to design the type taxonomy.

Section 3 — Content quality (60 minutes)

How to score each page

Pick your 10 most important pages (homepage, product, pricing, top blog posts). For each, run the editorial quotability checklist from "How to write content that gets cited by AI":

Score each page 0–10. Pages scoring 7+ are AEO-ready. Pages scoring 4 and below are the work.

Section 4 — Citation graph (45 minutes)

Search the open web for mentions of your brand. Specifically:

• Google: "<your brand>" -site:<your-domain.com>
• Reddit: site:reddit.com "<your brand>"
• Wikipedia: search your category page; do you appear? Should you?
• G2 / Capterra: do you have a profile? Is it complete? Are reviews recent?
• Industry pubs: search the top 5 in your niche

Tally:

Reading the citation table

If most rows are zero, your AEO ceiling is low — you can have perfect on-domain content and still lose because no third party has named you. The fix is the citation flywheel: directories first, niche industry publications next, mainstream press and Wikipedia (if you qualify on notability) last. None of these are quick wins, but each placement compounds for years.

Section 5 — Measurement (30 minutes)

Pick 30 prompts a real buyer in your category would ask. Mix branded, comparison, and category. Run them through ChatGPT, Perplexity, and Google AI Overviews. For each prompt, log:

• Were you mentioned? (yes/no)
• Was a competitor mentioned? (which one)
• What URLs were cited?
• What sentiment was attached to your mention?

This is your baseline. Without it, you can't measure whether your fixes are working.

For a more detailed framework, see Measuring share of voice in AI answers.

Putting it together: the audit deliverable

A useful audit produces a single-page output with:

1. Section scores (Crawler / Schema / Content / Citations / Measurement)
2. Top 10 fixes ranked by impact × effort
3. The 30-prompt baseline table
4. A 30-day plan that closes the highest-leverage gaps

Anything more is overkill. Anything less is not actionable.

Who should run this

This is one-marketer work. You don't need a dedicated SEO team. You don't need consultants. You need an afternoon, a JSON-LD validator, and the discipline to score honestly.

Honest scoring is the whole game

Most teams give themselves a pass on pages they wrote and grade competitor pages harshly. Force yourself to score each page as if you were a stranger hitting it for the first time. The discomfort of marking your own homepage 4/10 is exactly the discomfort that produces a useful audit. If everyone on the team scores their own pages above 7, the audit isn't an audit — it's a status report.

A worked example: scoring three pages

To make the framework concrete, here's how a typical first-time audit scores three pages from a hypothetical mid-market SaaS site.

Homepage. Crawler access: pass. Schema: missing — no Organization block at all. Content quality: 5/10 — strong hero copy, but no summary, no FAQ, no specific numbers, no Last updated. Citations: pass — five inbound mentions across industry blogs. Verdict: schema is the highest-leverage fix; ship in an hour.

Pricing page. Crawler access: pass. Schema: partial — SoftwareApplication exists but offers is a single object, not an array, so only one tier is exposed. Content quality: 6/10 — tiers and prices clear, but no comparison table, no FAQ, no testimonial schema. Citations: weak — no third-party "best pricing" pieces mention the product. Verdict: turn offers into an array, add an FAQ block with FAQPage schema, brief one comparison piece.

Top blog post. Crawler access: pass. Schema: Article present, but dateModified matches datePublished from 2023. Content quality: 4/10 — long flowery intro, no summary, single H2, no list or table. Citations: zero — no external links to this URL. Verdict: full rewrite against the editorial checklist, then a refresh push.

This pattern repeats across most audits. The wins are obvious once you score honestly: schema first (cheap and fast), then editorial (slower but compounding), then citation (slowest but most defensible).

After the audit: the 90-day plan

A useful audit feeds directly into a 90-day plan. Roughly:

• Days 1–14: ship every quick schema and robots.txt fix. These are 1–4 hour tasks each.
• Days 15–45: rewrite the top 10 pages against the editorial checklist. This is your biggest sustained effort.
• Days 46–75: publish 3–5 new cluster pieces aimed at the prompts you currently lose.
• Days 76–90: run the second baseline measurement. Compare to day 1. Adjust priorities.

Most teams will see a measurable share-of-voice change between baselines. If you don't, the issue is almost always citation graph: you've fixed the on-domain side but no third party has named you yet.

The bottom line

A serious AEO audit is finite and reproducible. Run it once a quarter. Pair it with a weekly measurement loop. Your visibility compounds linearly with how diligently you close the audit's gaps.

Start a free AEO scan — we'll surface the highest-leverage fixes for your domain in minutes.