Generative engine optimization is one of the few terms in this category that started somewhere respectable. Not an agency blog and not a product launch, but a paper with a benchmark and a reproducible result.
Most of what is written about it now has drifted a long way from that. This post covers what GEO stands for, what the original work actually claims, what holds up against what the engines themselves document, and which parts of the current GEO industry are selling something that does not exist.
What does GEO stand for?
GEO stands for generative engine optimization: optimising content so that generative engines, meaning assistants that synthesise one answer from several sources, are more likely to include it and cite it. ChatGPT, Claude, Gemini, Perplexity and Google's AI Overviews and AI Mode are all generative engines in that sense.
In search marketing, GEO has nothing to do with geography or geotargeting. It is about how often an AI answer includes you, and that splits into two separate outcomes: being named in the prose, and being cited as a source.
Where GEO comes from: the 2023 research paper
GEO was formalised in a 2023 paper by Aggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan and Deshpande, later presented at KDD 2024. The authors define generative engines as systems that answer a query by synthesising information from multiple sources and summarising it, and they make the point that this leaves content creators with "little to no control over when and how their content is displayed."
Their contribution is two things. A framework for optimising content against those engines, and GEO-bench, a benchmark of queries across domains with the web sources needed to answer them, so that claims can be tested rather than asserted.
How GEO works: an answer is not a ranking
Three things set it apart from ordinary search, and they are the reason the term earns its existence.
The outcome is not a position. A ranking puts you somewhere in a list. A generative engine includes you in a synthesised answer or it does not, and inclusion splits into two independent events: being named in the prose, and being cited as a source. A page can be cited while a competitor is named, which is its own problem with its own fix.
The retrieval set is wider than the query. Google generates related queries alongside the one typed and assembles from that broader pool, which it has described while expanding AI Mode. A page that answers one narrow question thoroughly can surface for an adjacent one.
The answer is generated, not retrieved. This is the part that breaks conventional measurement. Two runs of the same prompt an hour apart can name different brands with nothing having changed. Everything about how you verify a GEO change follows from that, and most published case studies ignore it.
None of this replaces SEO. Google describes its generative features as rooted in the core Search ranking and quality systems, so being retrievable in ordinary Search is still the precondition.
GEO strategy: what the research found works
The paper's most useful result is narrower than its headline. On its own benchmark, it reports that edits which give an engine something to cite, such as citing sources, adding quotations and adding statistics, did more for visibility than its other methods, while keyword stuffing, the reflex carried over from old SEO, did little.
That is a result on a controlled benchmark rather than a promise about your category, but it points the same way as what the engines themselves document. A GEO strategy, in order:
- Give the engine something to cite. A sourced figure, a quotation, a specific claim that can be checked. This is the part of the paper most worth taking literally.
- Be one of the sources worth synthesising. Specific rather than promotional. "Six engines on every plan" survives synthesis. "Industry-leading AI visibility" does not.
- Get described accurately where the model reads. The pages cited in your category are mostly not yours, so the roundups and comparison posts brief the model before your site does.
- Check that the engines can reach you. Audit robots.txt against each assistant's documented crawlers, and audit the snippet directives that restrict AI features. While you are in there, consider an llms.txt file, which is cheap hygiene rather than the ranking lever some vendors sell it as.
- Measure it properly. Which is the part everyone skips.
There is no GEO markup to buy. Google states that structured data is not required for generative AI search and there is no special schema.org markup you need to add. The formatting side of the work, question headings and passages that stand on their own, is the same as AEO's, and the checklist for it is in what AEO is.
GEO on ChatGPT, Claude, Gemini and Perplexity
A single "GEO score" assumes one target. There are at least six, and they behave differently.
OpenAI documents its crawlers separately by purpose, so blocking the training crawler and the search crawler are different decisions. Anthropic added web search to Claude in 2025, and in our measurements Claude cites fewer sources and is more willing to decline. Perplexity documents PerplexityBot and cites more heavily than anything else, which makes it the best early read on whether retrieval is working. Google runs AI Overviews and AI Mode as separate surfaces that disagree with each other on the same day.
Optimising for "AI" in the singular is how you end up strong on one surface and absent on another without knowing which. We report each engine on its own for that reason.
LLMO, LLM SEO and AI search optimization: other names for GEO
GEO has more names than techniques. LLMO (large language model optimization), LLM SEO and AI search optimization are all used for the same work: getting a brand included, named and cited in answers that a language model writes. AEO, answer engine optimization, is the closest relative, with the emphasis on being the answer rather than one of its sources.
The work under every one of those names overlaps almost completely, so choosing between them matters far less than deciding what you will measure. If a vendor's pitch turns on which acronym they use, that is a signal about the vendor rather than about your strategy.
GEO statistics: what the 40% figure does and does not mean
The headline result is that the paper's methods boost visibility "by up to 40% in generative engine responses" on its benchmark.
Read that carefully, because it is quoted badly everywhere. It is an upper bound, on a specific benchmark, using the paper's own visibility metrics. It is a genuine finding. It is not a percentage any vendor can promise you on your category, and anyone repeating it as one has not read past the abstract.
GEO-bench is a benchmark of queries with source documents. It is not a live measurement of ChatGPT, Gemini or AI Overviews as those products behave today, and the paper does not claim it is. Generative engines have changed substantially since 2023, including Google shipping AI Mode as a separate surface with its own retrieval.
So the honest reading is that the paper establishes the problem is real, tractable and measurable, and that content changes can move visibility meaningfully on a controlled benchmark. It does not establish which specific tactic will work on your category on a given engine this quarter. Only running the prompts will tell you that, which is the unglamorous conclusion this whole field keeps arriving at.
How to measure GEO
If the answer is generated fresh each time, a before-and-after screenshot is worthless. So is a single check, in either direction.
What produces evidence is a fixed prompt set, run on a schedule, with the wording held constant, on each engine separately, with every answer stored in full alongside its citations. Every rate carries the count behind it. Below five checks you show a fraction, because "2 of 4" is honest and "50%" implies a precision four runs cannot support. That is the substance of our methodology and it is deliberately boring. It is also all a GEO or AEO tool can honestly promise: the measurement, not the ranking.
Thirty or so runs per prompt per engine is roughly where a real change separates itself from ordinary variance. Changing the wording of a tracked prompt restarts the series, because a trend measured across two different prompts is not a trend.
For the definition sitting underneath all of this, see what AI visibility is. For how the assistants differ in practice, how AI chooses which brands to recommend. And if you want it without building it, RankHound is the AI visibility platform brands use to get recommended by AI. Every plan runs all six engines, keeps every answer, and shows whether a change you made worked.
GEO is a real discipline with a real paper behind it. Most of what is sold under the name is the measurement problem wearing a costume.
