AEO vs SEO: how AI Overviews pick a sentence

AEO (answer engine optimization) and GEO (generative engine optimization) describe the same job as SEO, pointed at AI Overviews, AI Mode, ChatGPT, and Perplexity. Google's own guidance is blunt: optimizing for generative AI search is still optimizing for the search experience.

If a page cannot rank, it is rarely cited. If a page ranks but never answers the question in a quotable block, models skip it for a competitor that did.

What is the difference between AEO and SEO?

SEO decides whether a URL is eligible. AEO decides whether a passage on that URL is extractable. You still need crawlable HTML, a clear title, and links. You also need a first-section answer that can stand alone.

A useful distinction:

• SEO asks whether Googlebot can index the page and match the query
• AEO asks whether a model can lift one or two sentences without the surrounding marketing
• GEO is the same work under another name, plus brand mentions on sites models already read

Do not treat them as separate roadmaps. A site that fails crawlability will fail both.

How AI Overviews usually choose a source

Third-party studies keep repeating the same pattern: most AI Overview citations come from pages that already sit on page one. A smaller share comes from deeper results when the passage is unusually clear.

That means the audit order stays conventional:

1. Can the crawler fetch the HTML without a client-only render?
2. Does the page target one query, not five products?
3. Is the direct answer in the first 40–60 words of the relevant section?
4. Are claims specific enough to quote (numbers, dates, named sources)?
5. Is the author or organization visible?

Googlebot renders JavaScript. Many other AI crawlers do not. If the answer lives only in a client component, ChatGPT and Perplexity may never see it even when Google can.

Siteory's AI search audit looks at that gap: rendered headings, answer-shaped passages, identity, and whether the HTML a bot downloads actually contains the copy.

Write passages models can cite

A citable paragraph answers the question in the first sentence, then adds one proof point. It does not open with "In today's digital landscape."

Compare:

• Weak: "There are many factors involved in appearing in AI Overviews, and it depends."
• Strong: "AI Overviews usually cite pages that already rank for the query and that answer it in a self-contained paragraph."

Use question headings when they match real queries. Keep paragraphs to two or four sentences. Put definitions in the "X is..." shape. Attribute numbers to a source and a date, or leave the number out.

Recency matters more for AI citations than most teams expect. A page that has not been reviewed in six months is easier to skip. Update the date only when you actually change the substance.

Brand mentions are not a backlink program

Ahrefs has published work showing brand mentions on YouTube, Reddit, and Wikipedia correlate more strongly with AI citations than raw Domain Rating. That does not mean you should farm mentions. It means a product with no public entity is hard for a model to name.

Practical work:

• Use one spelling of the product name everywhere
• Put Organization JSON-LD on the homepage with a real logo URL
• Write an about page that states who ships the product
• Earn mentions by publishing pages people can cite, not by posting the brand in threads

llms.txt does not replace any of this. See what llms.txt actually does.

What not to do for AI search

Skip these because they contradict Google's published AI optimization guidance or waste a sprint:

• Rewriting the same article "for ChatGPT" in a second URL
• Adding FAQPage schema after May 2026 to chase a rich result
• Blocking GPTBot while hoping to rank in ChatGPT
• Treating llms.txt as a ranking factor
• Chunking every page into 150-word cards because a slide deck said so

Do the opposite of blocking: keep public docs in HTML, allow the major AI user agents in robots.txt, and make the first answer on each important page boringly clear.

How Siteory scores AI search visibility

The AEO pass in Siteory is not a promise that ChatGPT will cite you. It reports what a crawler can prove from the HTML:

• Whether key answers exist as text, not only as images
• Whether headings look like questions a user would ask
• Whether the site identifies the organization
• Whether robots.txt or login walls hide the page from AI crawlers
• Whether llms.txt appeared in the crawl, reported as an observation, not a ranking claim

Run the AI search audit after the SEO audit. If the SEO pass is red, fix crawl and titles first. If the SEO pass is clean and the AEO pass is thin, the next edit is the opening paragraph, not another tool.