AEO / AI-Search Visibility Audit Prompt
insighttech.blog/library

Paste a page's URL and raw HTML (or just the URL if your model can browse)
into a chat model along with the prompt below. Works with any general-
purpose model — this doesn't require API access to a search index.

---

PROMPT:

You are an AEO (Answer Engine Optimization) auditor. I'm going to give you
a URL and/or the HTML of a page. Audit it against how AI answer engines
(ChatGPT browsing, Perplexity, Google AI Overviews, Claude) select and
cite sources. Go section by section:

1. STRUCTURED DATA — Is there valid schema.org markup (Article,
   Organization, FAQPage, HowTo, etc.)? Quote the @type values found.
   Flag anything malformed or missing that fits the content.

2. ANSWERABILITY — Find the single sentence or short passage on the page
   that most directly answers the page's core question. If no sentence
   could be lifted cleanly as a standalone answer, say so and explain
   why (buried in a listicle, requires scrolling context, etc.).

3. HEADING STRUCTURE — Is there exactly one H1? Do H2/H3s form a logical,
   skimmable hierarchy an AI crawler could use as a table of contents?

4. FRESHNESS SIGNALS — Is there a visible published or last-updated date?
   Is it in the visible HTML (not just metadata) where a model summarizing
   the page would see it?

5. CRAWLABILITY — Based on what's inferable from the page, would this
   content likely be blocked by a restrictive robots.txt for AI crawlers
   (GPTBot, ClaudeBot, PerplexityBot)? Note you can't fetch robots.txt
   directly — tell me to check with a tool like insighttech.blog/tools.

6. CITATION-WORTHINESS — Rate 1-5: if an AI answer engine were choosing
   between this page and 3 competitors for a citation, how likely is it
   to get picked, and why? Be specific, not just "seems fine."

Finish with a prioritized checklist: 3-5 concrete changes, ordered by
expected impact, not just a list of everything technically imperfect.

---

Notes:
- This is a manual/LLM-assisted audit, not a replacement for the
  rule-based checker at insighttech.blog/tools — use the checker for a
  fast automated pass, use this prompt for the judgment calls a rule
  engine can't make (answerability, citation-worthiness).
- Re-run this after each content refresh; freshness and answerability
  both drift as a page ages or gets edited by multiple people.