AEO stands for answer engine optimization. It is a young term, it is used loosely, and a fair amount of what is sold under it is ordinary SEO with a new label. That is worth saying first, because the useful part of the idea survives the skepticism and the rest does not.

The useful part is this. Classic SEO assumes a search engine returns a list of links and a person chooses one. Almost everything in the discipline follows from that assumption: you compete for a position, the position earns a click, the click is the outcome you measure. An answer engine — a chat assistant, an AI overview above the results, an agent doing a task on someone's behalf — does something different. It reads a set of pages and composes a response. Sometimes it cites the pages. Often the person never visits any of them.

The shift is from being ranked to being used

If the machine reading your page is going to summarise it rather than hand it over, the questions change:

  • Can it get the page at all? Answer engines fetch with their own user agents, from their own address space, and a WAF rule or a robots.txt line written years ago for scrapers may be refusing them. This is the single most common way a site is invisible to an assistant: not outranked, just not fetched.
  • Does the page say anything without JavaScript? Many fetchers do not run scripts. A page whose content arrives after hydration is, to those readers, an empty document with a nice title.
  • Is the claim you want quoted actually on the page, in plain words? A summariser can only repeat what it can find. Copy that gestures at a fact without stating it does not survive summarisation.
  • Is the page's own description of itself accurate? Titles, descriptions and structured data are how a page tells a machine what it is. They matter more, not less, when the machine is writing prose about you.

None of that is exotic. Most of it is what a careful SEO practitioner already did. The difference is which failures are now expensive. Ten years ago, a page that rendered only in a browser still ranked, because the dominant crawler executed JavaScript. Today the same page may be perfectly ranked and completely unquotable, because the reader that mattered for a given query never ran a script.

What AEO does not change

It does not replace SEO, and anyone telling you it does is selling something. The classic crawlers still index, the classic results still send traffic, and the two disciplines share most of their work: the same sitemap, the same canonicals, the same page speed, the same "does this page deserve to exist" question. Treating AEO as a separate channel with its own budget usually means paying twice for one set of fixes.

It also does not give you a rank to chase. There is no position four in an assistant's answer. There is cited, mentioned, or absent, and the boundary moves per query, per model, and per week. If a tool offers you a stable AEO score, ask what it is measuring — the honest answer is usually "a checklist of page properties", which is useful, and is not a ranking.

What can actually be measured

Less than the category's marketing implies, and more than nothing.

You can measure the mechanical side directly: whether a given fetcher gets a 200 or a block, whether the content is in the HTML, whether your declared metadata matches the delivered page. Those are facts about your own server and they are checkable on demand.

You can measure arrivals. When a person clicks through from an assistant to your site, the referrer usually says so, and that is a real number about a real visit. In our own analytics we separate four kinds of hit — crawler, search, referral and agent — because folding them together produces a figure that means nothing: a crawl is not a visit, and a visit from a chat interface is not the same event as a bot fetching the page to build an index.

What you generally cannot measure is the answer itself. Unless a model surfaces a citation you can see, you do not know whether your page informed a response, and you certainly cannot count how often. Anybody quoting you a precise share-of-voice figure across assistants is estimating, and the estimate deserves the same scrutiny you would give any other survey.

A reasonable working definition

AEO is the part of technical and editorial SEO that assumes your reader is a machine writing a summary for a person, rather than a person choosing a link. It changes emphasis rather than method, in four places:

  • access before ranking;
  • plain text before presentation;
  • stated claims before implied ones;
  • and honest metadata, because metadata is now testimony rather than decoration.

If you want a single first action, it is not a schema markup project. It is to check what an AI fetcher actually receives when it requests your most important page — status code, headers, and the HTML before any script runs. A surprising number of sites discover the answer is a block page, and no amount of optimization downstream of that matters.