AI search: getting cited by ChatGPT and Perplexity

A growing share of the questions people used to type into Google now get answered by something that reads the web and replies in a paragraph. Nobody clicks through. If your business appears in that paragraph you exist; if it doesn't, you don't — and the ranking that used to protect you doesn't apply.

This section covers what actually influences that. How answer engines decide which sources to cite, what structured data does for a machine that is reading rather than crawling, and what llms.txt is for — a file that is neither magic nor useless, and it's worth knowing which of the two it is before someone sells it to you as a service.

Also the terminology, briefly, because SEO, AEO and GEO get used interchangeably and they are not the same job. They overlap enough that most of the work counts twice, and differ enough that optimising for one and assuming the other follows is how sites end up with impressions that never turn into visits.

Which is the honest caveat this whole section deserves: being cited by an AI produces a mention, not necessarily a click. It is real, it brings real clients, and it is very hard to measure with the tools most businesses have installed. Worth doing. Worth measuring properly before you decide it worked.

If you only take one practical thing from this section, take this: the pages that get cited are the ones that answer a specific question with a specific number, in plain language, near the top. Not the longest pages, and not the ones stuffed with keywords. That is also, conveniently, what makes a page useful to a person who does click.