What is generative engine optimisation (GEO)?

Generative engine optimisation, or GEO, is the practice of making a website's content more likely to be used and cited when an AI system writes an answer. The systems in question include ChatGPT, Google's AI Overviews and AI Mode, and Perplexity. People searching for "GEO SEO" are usually asking whether this is a new discipline or a new name for an old one. The short answer, supported by Google's own documentation, is that it is SEO applied to a new kind of results page.

Where the term comes from

The term was introduced in a research paper, GEO: Generative Engine Optimization, first posted in November 2023 by researchers from Princeton University and the Indian Institute of Technology Delhi, with two independent co-authors. It was published at the KDD conference in August 2024.

The paper describes a generative engine as a search system that gathers information from several sources and uses a language model to write one answer, with the sources cited. It points out the problem this creates for website owners: they have little control over whether or how their content appears in that answer. GEO is the authors' name for the work of improving that visibility.

How a generative engine uses a website

Google's documentation explains the mechanism in two terms.

• Grounding, also called retrieval-augmented generation. The system pulls relevant pages from the search index, reads them and writes a response supported by links to those pages.
• Query fan-out. The system turns one question into several related searches to collect a wider set of pages. Google's example is a question about a weedy lawn producing searches such as "best herbicides for lawns" and "remove weeds without chemicals".

The consequence is simple. A page can only be used in an AI answer if the search system behind it can find the page first. That is why GEO and SEO overlap so heavily.

What the research found

The GEO paper tested nine ways of changing a page's content and measured how much of that page then appeared in generated answers. According to the paper:

• Adding quotations, adding statistics, improving fluency and citing sources were the most effective methods.
• Keyword stuffing performed worse than making no change at all.
• The best methods improved the paper's main visibility measure by about 40%.
• Results varied by subject area.

These figures need context. The main experiments ran on a simulated engine built by the researchers, which used the top five Google results and an older OpenAI model. The methods were also tested on Perplexity. They were not tested on today's ChatGPT or on Google's AI features, and the scores describe how much of a source appears in an answer, not how many customers it brings. The useful finding is the direction: content with specific, verifiable substance did better than content written to repeat keywords.

What Google says about GEO

Google addresses the term directly in its guide to optimising for generative AI features, last updated in July 2026. It defines GEO as generative engine optimisation and AEO as answer engine optimisation, and states that optimising for generative AI search is optimising for the search experience. In Google's view, that makes it SEO.

The same guide lists things site owners do not need to do for Google's AI features:

• Create llms.txt files or other special machine-readable files
• Break content into small chunks
• Rewrite content in special phrasing for AI systems
• Seek out inauthentic mentions on other sites
• Add special structured data

It recommends content that draws on real knowledge and experience, a sound technical structure, and accurate Google Business Profile and Merchant Center details.

Google's statements cover Google's products only. They do not describe how ChatGPT or Perplexity work.

GEO, AEO and AI SEO

These terms are used loosely and often for the same work.

• GEO comes from the research paper and refers to engines that generate a written answer from several sources.
• AEO, answer engine optimisation, is an older term for getting content chosen as the direct answer to a question, for example in a featured snippet or a voice assistant reply.
• AI SEO and AI search optimisation are general labels that cover both.

In practice the tasks are the same. The label matters less than whether the work is sound. Google's guide advises checking any third-party GEO or AEO advice against its own guidance.

What GEO changes in practice

Stryx's position is that core SEO comes first: pages built correctly, one topic per page, and content that says something specific. Two additions then help both readers and AI systems:

• A short summary at the top of each page that answers the main question directly
• Clear questions and answers on the page

What does change is measurement. An AI answer can mention a business without sending a visit, so clicks alone understate visibility. Google's Search Console now reports impressions in its AI features separately.

This article defines the idea. The practical steps are in the guides to getting recommended by ChatGPT and Google AI Overviews and AI Mode, and the comparison of GEO and SEO sets the two side by side. Stryx offers this work as AI search optimisation. To discuss a site, get in touch.