AEO SEO checklist: what to fix before AI answers can quote you

·7 min read

AEO SEO is the work that decides whether a search engine or AI assistant can lift a passage from your page and use it as the answer. For a single page it comes down to five checks, run in order. The page has to be fetchable and indexed, the answer has to be easy to find, the passage has to make sense when quoted alone, your snippet rules have to allow quoting, and your structured data has to match what the page shows.

Every item below has a test you can run in under a minute with curl or view-source. If the term is new, start with what answer engine optimization is. This list covers one page. Domain-wide crawler access and off-site brand work belong in the GEO checklist.

How to use this AEO SEO checklist

Start with the pages closest to being quoted, usually the ones that already rank for question queries. Since August 31, 2026, every site can also see which pages earn impressions in AI Overviews and AI Mode in Search Console's Generative AI performance report. It counts impressions only, with both features combined, so use it to pick candidates.

  1. Write down the question the page should answer, in the words a searcher would type.
  2. Run the fetch and index tests first. A page with a noindex tag gets nothing from better headings.
  3. Work down the other groups in order, then rerun the list after you deploy, because one template change can undo a fix on every page that shares it.

Can the page be fetched and indexed?

It has to be, because Google only uses a page as a supporting link in AI Overviews or AI Mode when it is "indexed and eligible to be shown in Google Search with a snippet." Its AI features documentation lists no other technical requirement.

Check One-minute test
The final URL returns 200 First command below. Anything but 200 after redirects fails.
No noindex Search view-source for noindex, then run the second command.
robots.txt allows the crawlers you want Find the group in /robots.txt that names each bot.
The canonical points at this URL Search view-source for rel="canonical".
The answer is in the raw HTML Third command, with the first words of your answer.
# Status and final URL after redirects
curl -s -o /dev/null -L -w "%{http_code} %{url_effective}\n" https://example.com/what-is-a-canonical-tag

# Robots rules sent as headers, on every hop
curl -sIL https://example.com/what-is-a-canonical-tag | grep -i x-robots-tag

# 0 means the sentence is not in the HTML the server sends
curl -s https://example.com/what-is-a-canonical-tag | grep -c "A canonical tag is"

In robots.txt, a named group replaces the * group. Under RFC 9309, a crawler follows the group that names it and falls back to * only when none does, so a friendly User-agent: * block does nothing for PerplexityBot if a PerplexityBot group says Disallow: /.

Search crawlers and training crawlers are also different bots. OpenAI's crawler documentation says a site that blocks OAI-SearchBot won't appear in ChatGPT search answers, while GPTBot collects content that may be used to train OpenAI's foundation models. Perplexity says PerplexityBot surfaces and links sites in its search results and is not used to crawl for foundation models. This file keeps the training crawler out and lets the search crawlers in.

User-agent: GPTBot
Disallow: /

User-agent: OAI-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: *
Allow: /

Google processes JavaScript as long as the scripts aren't blocked, per its AI optimization guide. Vercel's study of AI crawler traffic, published in December 2024, found that none of the major AI crawlers, GPTBot, ClaudeBot and PerplexityBot among them, executed JavaScript. If the grep returns 0, assume those crawlers see a page with no answer on it.

Can an engine find the answer on the page?

It can when each heading asks a question in the searcher's words and the first sentence under it answers that question. Google's AI features run what Google calls query fan-out, issuing several related searches across subtopics to build one response, and each of those searches is narrower than the question the user typed. "How long does a canonical change take to show up?" tells the engine what a section answers. "Timing" does not.

Check One-minute test
Headings use the searcher's words Read only the H2s and H3s. Each should work as a search query.
The first sentence answers the heading Read it on its own. If it needs the next sentence, it fails.
One question per section A section that answers two questions needs two headings.
Headings are real heading elements Search view-source for <h2. A bold paragraph is not a heading to a parser.

Google's AI optimization guide calls semantic HTML generally a good idea, and real heading tags are the cheapest part of it. For how to rewrite sections that fail, read how to optimize content for AI search.

Can the passage be quoted on its own?

A passage can be quoted on its own when it still makes sense, and still says something specific, after it is cut away from the page around it. Paste one section into a blank document and read it cold. If you can't tell which product, place or version it is about, put the name back in.

Check One-minute test
No backward references Search for "as mentioned", "above" and "earlier". Check the section doesn't open with "It" or "This".
Specific numbers Replace "a few days" with the real range, such as "3 to 7 days".
Sourced facts Every statistic links to the organization that published it.
Comparisons and steps as text Search view-source for <table and <ol. A comparison saved as an image has no text to quote.

Don't take this as permission to chop the page into fragments. Google's AI optimization guide says there is no requirement to break content into tiny pieces, and that optimizing for generative AI search is still SEO. You want sections that stand alone, whatever their length.

Do nosnippet and max-snippet stop AI answers from quoting you?

Yes. In Google, nosnippet and max-snippet apply to AI Overviews and AI Mode as well as regular results, so an old rule meant to hide a snippet also stops Google from using the text as direct input for AI answers, or caps how much it can use. The details are in Google's robots meta tag documentation.

Rule Effect in Google
nosnippet or max-snippet:0 No text snippet, and no use as direct input for AI Overviews or AI Mode.
max-snippet:50 Caps the snippet at 50 characters and limits how much text AI Overviews and AI Mode can use.
max-snippet:-1 No limit. Google chooses the length.
data-nosnippet Hides part of a page from snippets. Works only on span, div and section elements.

Search view-source for both rules, then run the second curl command, because they can also arrive in an X-Robots-Tag header that view-source never shows. To hide one line, scope it instead of limiting the whole page.

<!-- Let Google choose the snippet length -->
<meta name="robots" content="max-snippet:-1">

<!-- Hide one sentence, keep the rest quotable -->
<p>Shipping takes 3 to 5 business days. <span data-nosnippet>Use code SPRING10 at checkout.</span></p>

Bing uses different rules. When it announced them in September 2023, Bing said NOARCHIVE content is left out of Bing Chat answers, while NOCACHE content can appear only through its URL, title and snippet.

Google also has a sitewide switch. The Search generative AI control in Search Console settings reached all sites on August 31, 2026 and defaults to include. Exclude keeps the site out of AI Overviews, AI Mode and Discover's generative AI features. Check it once, because no page-level fix gets an excluded site into Google's AI answers.

Which structured data still matters for AEO?

Markup that accurately describes what the page visibly shows still matters, but none of it is required to appear in AI answers. Google's AI optimization guide says structured data isn't required for generative AI search, though Google still recommends it for rich results.

FAQ markup is where most checklists are out of date. Google limited FAQ rich results to well-known government and health sites in August 2023 and stopped showing them for every site on May 7, 2026, according to the Search Central changelog. A real, visible FAQ can keep its markup. Adding an FAQ block only to carry markup no longer buys a Google rich result.

Check One-minute test
JSON-LD parses Paste the URL into Google's Rich Results Test.
Values match the visible page Google's structured data policies forbid marking up content readers can't see. Compare headline, price, author and dates.
Dates agree dateModified matches the updated date printed on the page.
Authors are identifiable Article author has a url to a real bio page, a recommended property in Google's Article markup documentation.
No FAQPage added for rich results If the FAQ exists only for the markup, cut both.

Check a page against the AEO checklist

Each group maps to one of our tools, priced per run in credits.

Checklist group Tool Credits per run
Fetchable and indexable, snippet rules Indexing & Canonical Checker 8
Answer in the raw HTML AI Crawler View 8
AI crawlers allowed in robots.txt Robots.txt & AI Crawler Checker 10
Answer can be found Question Finder 15
Passage can be quoted Citability Analyzer 5
Structured data Schema Markup Checker 8

The Indexing & Canonical Checker reports the status code, noindex, the canonical and snippet rules, including data-nosnippet on elements Google ignores, and names the tag or header to change. AI Crawler View lists content and headings that appear only after JavaScript runs. The Citability Analyzer flags sections that don't open with an answer, lean on text outside them or run past 400 words.

Question Finder pulls questions from Google autocomplete and marks each one as answered, given a heading but no direct answer, mentioned or missing on your page. The Schema Markup Checker lists missing required and recommended properties and flags FAQPage markup that no longer earns a Google rich result. Both are on the AEO tools page, along with the schema builders that generate the markup.

Keep reading