Generative Engine Optimization (GEO)
Generative Engine Optimization (GEO) is the practice of influencing what a generative model says about a topic, brand, or person — through the sources it retrieves and the training corpus it absorbed. AEO targets a single extractable answer; GEO targets the model's overall representation of you.
GEO is a reputation problem wearing a technical costume
When someone asks a model 'who are the leading thinkers on incentive design?', the model is not ranking pages. It is summarizing a distribution of statements it has seen across the web. The lever is corpus presence: how consistently an entity is described the same way, in enough independent places, with enough specificity to survive compression.
What moves the representation
Consistency is the highest-leverage variable. A single canonical description of an entity, repeated across owned pages, third-party profiles, podcasts, transcripts, and structured data, compresses into a stable answer. Contradictory or vague descriptions compress into nothing.
- —One canonical entity description reused everywhere, including sameAs links.
- —Third-party corroboration: interviews, transcripts, citations, coverage.
- —Original artifacts — frameworks, named methods, datasets — that give the model a handle.
- —Depth on a narrow territory rather than shallow coverage of many.
How to audit your current representation
Ask five engines the same eight questions about your category, your company, and your named experts. Record the exact wording of the descriptions that come back. You are looking for three failure modes: the model does not know you, the model knows you but describes you as something adjacent and wrong, or the model attributes your framework to someone else.
Each failure has a different fix. Absence is a corpus problem — you have not published enough resolvable material. Misdescription is a consistency problem — your own pages describe you three different ways. Misattribution is a priority problem — someone else published the same idea more clearly, more often, or earlier.
The canonical description test
Write one sentence that defines the entity, then check whether that exact sentence appears on the homepage, the about page, every author hub, the Organization schema, the LinkedIn profile, the podcast bio, and the speaker one-sheet. If the sentence drifts, the model has to average across variants, and averages compress into vagueness.
- —One sentence, under 30 words, containing the category and the differentiator.
- —Reused verbatim in prose and in structured data description fields.
- —Corroborated by at least three sources you do not own.
- —Anchored to a named method or artifact the model can point at.
What GEO cannot do
GEO cannot make a model say something the open web does not support, and attempts to force it — mass-generated pages, fabricated credentials, coordinated identical text — read as spam to the same deduplication systems that decide citations. The durable play is to be genuinely the clearest source on a narrow territory, then make that clarity machine-legible.
Frequently asked
- Is GEO the same as AEO?
- They overlap but differ in scope. AEO optimizes a page to be cited as the answer to a specific question. GEO optimizes the model's aggregate representation of an entity across every question.
- Can you 'rank' in a generative engine?
- There is no ranked list to occupy. What exists is citation frequency and representation accuracy — how often you are used as a source and how correctly you are described.
Related definitions
Answer Engine Optimization
Answer Engine Optimization (AEO) is the practice of structuring content so answer engines — ChatGPT, Google AI Overviews, Perplexity, Claude, and Copilot — can extract, cite, and reuse it as the answer to a question. Where SEO competes for a click on a ranked link, AEO competes to be the sentence the machine says out loud.
Entity SEO
Entity SEO is the practice of making the people, organizations, and concepts on a site unambiguously identifiable to machines — through consistent naming, structured data, and links to authoritative profiles. Search and answer engines reason over entities, not keywords, so an unresolved entity cannot be credited, cited, or trusted.
Citation Share
Citation share is the percentage of AI-generated answers to a defined set of questions in which a given domain appears as a cited source. It is the primary scoreboard for answer engine optimization, replacing keyword rank as the measure of visibility when the answer, not the link, is the destination.
llms.txt
llms.txt is a plain-text file at the root of a website that gives large language models a curated, machine-readable map of the site: what the site is, who stands behind it, and which URLs carry the canonical content. It plays the role robots.txt plays for crawlers, but for retrieval and comprehension rather than permission.