Most AEO advice is recycled SEO with a new label

Search the term and you will get the same ten recommendations from every article. Add schema markup. Use clear headings. Improve page speed. Write in a question and answer format. Build topical authority.

None of that is wrong. All of it is 2015 technical SEO, and the reason it keeps getting published as AI search strategy is that it is safe, familiar, and easy to sell as a deliverable. It also stops well short of the part that decides whether a model names you.

Ranking and being quoted are different jobs

A search engine ranks documents. It decides which page best matches a query, hands over ten of them, and lets the human choose. A language model constructs an answer. It decides which claims to state, then which sources support those claims well enough to be worth naming.

The three things that actually differ

1. 1

   Extractability

   A model lifts passages, not pages. If the answer to the question in your heading takes four paragraphs of preamble to arrive at, it will not be lifted. Put the complete answer in the first two sentences under the heading, then expand. This feels like bad blogging and it is exactly what gets quoted.

2. 2

   Verifiability

   Unsourced assertions are the cheapest thing on the internet and models treat them accordingly. A claim with a named study, a date and a number is safer to repeat than the same claim stated confidently. If your content contains no numbers you can defend, it will lose to content that does.

3. 3

   Consensus

   This is the one that has no SEO equivalent worth the name. Models synthesise across sources. If your site says you serve mid market fintech, a review platform says SMB retail, and three listicles put you in a different category entirely, the model resolves the conflict by describing you vaguely or skipping you. Off site agreement about your category is not brand hygiene, it is a ranking input.

You cannot optimise your way into an answer that a model does not trust enough to attribute.

Where schema actually earns its place

I am not arguing against structured data. It resolves ambiguity cheaply. It tells a parser which entity you are, when a page changed, who wrote it and what their credentials are. That matters, and it costs an afternoon.

What it does not do is make an unsupported claim credible or resolve a contradiction between your site and the rest of the web. Schema is punctuation. It clarifies a sentence that already says something. Most AEO advice sells punctuation as an argument.

A quick test for any AEO recommendation

The honest summary is that the technical layer is a one time cost and the credibility layer is the ongoing work. Teams keep buying the first because it has a clear finish line, and avoiding the second because it involves other people's websites, real research, and having something specific to say.