Ask ChatGPT which SEO plugin suits a WooCommerce store and you get an answer, not ten blue links. Google’s AI Overviews put an answer above the results, and Perplexity was built that way from the start. For a growing share of questions, the answer is where the search ends. The pages it quotes get the credit, and sometimes the click.
Generative Engine Optimization (GEO) is the name for making your site one of those pages. It sounds like a new discipline. Mostly, it is SEO done properly, plus a few habits that matter more when a machine does the reading.
Where the term comes from
GEO comes from a 2023 research paper by researchers from Princeton and other institutions, presented at the KDD conference in 2024. They measured how changes to a page affect how often AI engines use it in their answers. The biggest gains came from content that is easy to check: citing sources, adding quotations and adding statistics lifted a page’s visibility by up to 40%. Keyword stuffing did not help.
That is worth holding on to, because most of what gets sold as GEO today is not in that paper.
How AI answer engines pick their sources
The details differ per engine, but the pattern is the same. The engine turns a question into one or more searches, runs them against a search index, reads a handful of the pages that come back and writes an answer from what it read, usually with links to its sources.
That index belongs either to a search engine or to the AI company itself. Google’s AI Overviews draw on Google’s own index. ChatGPT search combines OpenAI’s crawler, OAI-SearchBot, with third-party search providers. Perplexity crawls with PerplexityBot. So the first rule of GEO is an old one: a page that isn’t indexed can’t be quoted.
What actually helps
Be indexable. A sitemap, canonical URLs and no noindex on pages that should be found. If you block crawlers in robots.txt, know which ones you block. GPTBot and ClaudeBot collect training data; OAI-SearchBot, PerplexityBot and Claude-SearchBot fetch pages for AI search. Block the search crawlers and your pages show up less, or not at all, in those answers. Google-Extended only governs whether Google may use your pages for Gemini; it has no effect on Google Search, and AI Overviews are part of Search.
Put the answer first. A heading that asks the question, and the answer in the sentence right below it. An engine that reads a handful of pages in a second can only quote what it can find, so don’t make it dig.
Be specific and checkable. Numbers, dates, prices, named sources. These are exactly what made pages more visible in the GEO research, and they are what a careful reader wants anyway.
Say what each page is. Structured data (schema.org JSON-LD) tells machines that a page is an article by a named author, a product with a price and stock, or the homepage of an organization. A meta description says in one sentence what the page is for.
Be worth quoting. Everything in why AI content alone won’t rank applies twice here. Thin pages written for bots don’t get quoted, by Google or by anyone else.
Where llms.txt fits
llms.txt is a proposed standard from 2024: a plain-text file at the root of a site with its name, a one-line summary and links to its most important pages, written in Markdown so a language model can read it without wading through HTML. AI tools and coding agents read it, and the AI labs publish one for their own developer documentation.
What it isn’t, yet, is a proven way into AI answers. Google says you don’t need new machine-readable files, AI text files or markup to appear in its AI features. And so far there’s no solid evidence that an llms.txt makes ChatGPT or Perplexity cite a site more often.
We would still add one. It takes minutes, costs nothing and gives any tool that does read it a clean summary of your site. Just don’t make it the plan.
How SEO & Chill handles GEO
The plugin does the groundwork on every page, so your time goes to the part only you can do: content worth quoting.
llms.txt and llms-full.txt, built from your pages and rebuilt when you publish. You write the one-line summary, or let your AI provider draft it.
Structured data as JSON-LD: your site and organization, your pages, articles with their author, products with price and stock, and breadcrumbs.
Meta titles and descriptions from the AI Meta Writer, written to say plainly what a page is for. One page at a time on Basic, in bulk from Pro.
XML sitemaps, listed in robots.txt, and canonical URLs, so the pages you care about stay in the index.
llms.txt, schema and sitemaps work on every plan. Schema, Sitemaps & llms.txt has the details, including what we don’t promise.
llms.txt, schema and sitemaps come with every plan of SEO & Chill.
The calm SEO letter. Three short tips to start, then one practical email a month, written by the team. No drama, unsubscribe anytime.