Where the term comes from
DOCUMENTED Generative Engine Optimisation is unusual among SEO terms in having a citable origin. The paper “GEO: Generative Engine Optimization” by Pranjal Aggarwal and colleagues was posted to arXiv on 16 November 2023 and accepted at KDD 2024. Its abstract defines the problem: large language models have created a new kind of search engine that “synthesizes information from multiple sources” and answers the query directly, and content creators “have little to no control over when and how their content is displayed”.
TESTED The authors introduced GEO as “a flexible black-box optimization framework for optimizing and defining visibility metrics”, built a benchmark of queries across domains (GEO-bench), applied a set of content changes to source documents, and measured how the changed documents fared in generated responses. Their headline result was that GEO “can boost visibility by up to 40% in generative engine responses”. Their own caveat follows immediately: “the efficacy of these strategies varies across domains, underscoring the need for domain-specific optimization methods”.
Two limits matter when that figure is quoted, and it is quoted often. It was measured on the authors’ benchmark and generative engine setup, not on any commercial product as it operates today. And it was a ceiling across strategies and domains, not a typical result. The paper supports the claim that content changes can change generative visibility; it does not support a claim that any given change will produce any given lift on ChatGPT or Google.
What GEO optimises
The paper’s visibility metrics counted how prominently a source appeared in a generated answer: whether it was cited, how much of the answer drew on it, and where it sat. That is the useful definition of the practice. GEO is about being a source the engine composes from, and being attributed for it, rather than holding a position in a list.
In practice the work is the same set of foundations this site describes under AI SEO: access for retrieval agents, an identifiable business, passages that answer the question directly, corroboration from sources the business does not control, and original material worth citing. The name signals emphasis on the synthesis step; it does not describe a different mechanism.
GEO, AEO and AI SEO
The three terms are used interchangeably in the market and it helps to be precise about the difference in emphasis, because it is only emphasis.
| Term | Origin | Emphasis | This site’s usage |
|---|---|---|---|
| AI SEO | Market term | The whole practice across AI-assisted search | Umbrella term |
| GEO | Academic (2023) | Being composed from and cited by generative engines | Research-origin synonym |
| AEO | Market term, predates generative AI | Being the direct answer on answer surfaces (featured snippets, voice, AI answers) | Sibling term for direct-answer surfaces |
INFERRED The distinction this publication draws: GEO is the right word when the question is about synthesised, multi-source answers with citations; AEO is the right word when the question is about being the single extracted answer. Most work serves both.
What Google says about optimising for its generative surfaces
DOCUMENTED Google’s site-owner documentation for AI Overviews and AI Mode says the existing best practices for SEO remain relevant, that there are “no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary”, and that a page must be indexed and eligible to appear with a snippet to be a supporting link. It also says both surfaces “may use a query fan-out technique”, issuing multiple related searches across subtopics to develop a response, and that AI Mode and AI Overviews “may use different models and techniques”.
That is the clearest first-party statement available on GEO for any major platform, and it cuts against most GEO sales copy: on Google, GEO is SEO done well, plus the understanding that a fan-out can retrieve a page for a sub-question the person never typed.
How GEO applies in Perth
DOCUMENTED Both Google generative surfaces are live in Australia. AI Overviews lists Australia among its available countries, and Google Australia announced AI Mode’s rollout “in Australia in English” on 8 October 2025, describing the same query fan-out behind it.
HYPOTHESIS Whether a Perth business is composed into those answers for Perth questions depends on which sources the engines retrieve for Perth sub-queries. That is the object of the planned Perth source studies on this site and is not asserted here. The Perth-specific point that can be made from the documentation is that fan-out rewards a site that covers the sub-questions around its subject, including the local ones, rather than one page aimed at one phrase.
What is documented and what is inferred
- Documented: the paper’s existence, authorship and definitions; Google’s no-special-optimisation statement; query fan-out; availability in Australia.
- Tested: the paper’s headline visibility result, on its own benchmark, varying by domain.
- Inferred: the emphasis-only distinction between GEO, AEO and AI SEO.
- Not established: any transfer of the paper’s percentages to a live platform, and any Perth frequency.