Generative Engine Optimisation (GEO)

The practice of structuring a website so AI answer engines (ChatGPT, Claude, Perplexity, Google AI Overviews) can ingest, ground, and cite its content reliably. Distinct from SEO, which optimises for blue-link ranking. GEO priorities include machine-readable content surfaces (llms.txt, JSON-LD), definition-shaped ledes, and a cross-linked entity graph.

Explore in

In depth

SEO optimises a page to win a blue link. GEO optimises a page to be the source an answer engine quotes. The two overlap. Both reward crawlable HTML, fast load times, clean semantics. They diverge at the top of the stack. SEO cares about click-through rate, dwell time, and keyword placement. GEO cares about whether an LLM retrieved the right page, whether it could parse the page into a self-contained answer, and whether the answer it produced included an attributable citation back.

Concretely, a GEO-ready site looks different. It ships an /llms.txt index so models have a curated map of the corpus. It exposes per-article /writing/<slug>/llms.txt clean-text surfaces so fetchers get content without page chrome. It emits rich JSON-LD (Article, DefinedTerm, FAQPage, Person) so the entity graph is explicit. Its ledes are short, declarative, quotable sentences an AI can lift into an Overview without paraphrasing. And it keeps a cross-linked glossary so terms resolve to stable URLs the model can cite.

The operator job has shifted accordingly. A site's GEO posture is now an architectural property, not a marketing tactic. It is decided in the schema, the routing, the build pipeline, and the editorial template, long before any individual article is written. Publishers who treat GEO as a content-team responsibility tend to under-deliver. Publishers who treat it as a site-architecture responsibility tend to compound, because every new article inherits the citability of the structure underneath.

Examples

  • A how-to article that opens with a 40-word standalone answer to the title's question, so that an AI Overview can quote the first paragraph verbatim and attribute it.
  • A site that publishes /llms.txt, /corpus.json, and per-article clean-text endpoints. Fetchers find the machine-readable entrypoints immediately; grounding quality improves; citation rate goes up.
  • A glossary where every term has its own URL, its own DefinedTerm JSON-LD node, and its own clean-text endpoint. Articles cross-link to glossary entries; glossary entries list the articles that reference them. The entity graph is machine-readable end-to-end, which is what makes any single answer citable rather than paraphrased.

Usage notes

GEO is additive to SEO, not a replacement. A page still needs to be crawlable, fast, and well-structured for blue-link discovery. The difference is the optimisation target. You're no longer only writing for a reader who clicks through. You're also writing for a model that will quote you without sending the reader at all.

Also known as

  • geo
  • generative engine optimisation
  • generative engine optimization

These aliases are what the site's build-time auto-linker matches against to cross-reference this term across the FAQ and machine-readable endpoints.

Referenced in

Related terms

  • AI-native publishing — A publishing operating model where AI agents handle research, drafting, editorial review, SEO/GEO, and programming as default, with human operators overseeing strategy and judgement calls.
  • The $0 CPM problem — The advertising-economics crisis for online publishers as AI chat interfaces replace search-driven traffic.
  • Trust as the last moat — Trust as the last moat is the claim that, in a world where any text or image can be generated cheaply and credibly, reader-facing trust becomes the only durable defensibility a publisher has.

← Back to the glossary