How AI answers pick sources
When an assistant answers with live web results, it runs searches, fetches a handful of pages, and writes an answer from what it read, often with links to the sources. A page is cited when it was retrieved and when it states the needed fact clearly enough to quote.
Assistants that answer from training data alone rely on what was in the crawl when the model was trained. In both cases, content that crawlers can fetch and models can parse has an advantage.
What GEO work looks like
Allow AI crawlers in robots.txt. Serve real HTML, not an empty shell that needs JavaScript. Put a plain one or two sentence answer under each question style heading. State facts with numbers and dates. Mark up entities with schema. Publish an llms.txt that points models at your most useful pages. Keep a consistent name and description everywhere your product is listed, so models connect the mentions.
Being mentioned on pages that already get cited matters too. Launch platforms, comparison pages and community threads are frequent sources for "what is the best tool for X" questions.
GEO and SEO
Most of GEO is good SEO: crawlable pages, clear structure, real expertise, and being referenced by trusted sites. The difference is emphasis. GEO rewards quotable sentences and factual density over keyword targeting, because the reader is a model writing a summary.
Example
A founder rewrites their homepage hero from "Supercharge your workflow" to "Acme is a time tracker for freelance designers that turns tracked hours into invoices." Assistants asked for an invoicing time tracker can now quote it.
Questions people ask
- How do I know if AI assistants cite my site?
- Ask them the questions your customers ask and see which sources they name. In analytics, watch for referrals from chatgpt.com, perplexity.ai, claude.ai and copilot.microsoft.com.
- Is GEO different from AEO?
- They overlap. AEO focuses on direct answers in search features and voice assistants. GEO focuses on generative AI answers. The techniques are largely shared.
Related terms
- Answer Engine Optimization (AEO)Answer Engine Optimization (AEO) is shaping content so that search features and assistants can lift a direct answer from it, such as a featured snippet, a People Also Ask box, a voice answer or an AI summary.
- llms.txtllms.txt is a Markdown file at the root of a website that gives large language models a short summary of the site and links to its most useful pages, so they can read it without parsing every page.
- Markdown twinA Markdown twin is a plain Markdown version of an HTML page, served at a parallel URL such as page.md or on request with Accept: text/markdown, so AI tools can read the content without the page's markup.
- ai.txtai.txt is a plain text file at a site's root that states, in plain language, how AI systems may use the site's content and which facts they should know about it.
- Schema markupSchema markup is structured data, usually written as JSON-LD, that describes a page's content in the schema.org vocabulary so search engines and AI systems can understand it without guessing.