- GEO stands for Generative Engine Optimization. It means working toward being mentioned and linked as a source in answers from ChatGPT, Gemini, Perplexity or Google's AI Overviews.
- GEO builds on SEO. Anything a bot cannot fetch, crawl and render is also missing from AI answers. Google itself says that optimizing for its AI search is still SEO.
- What is new is mainly the goal and the measurement. Instead of position and click, what counts is the mention, the recommendation and the cited source, and you measure those via prompts and logs.
What is GEO?
The term comes from research. Pranjal Aggarwal and colleagues published the paper “GEO: Generative Engine Optimization” in November 2023. It was accepted for the KDD 2024 conference. In it, the authors test how changes to content affect visibility in the answers of generative search systems. According to the abstract, visibility in their test environment rises by up to 40%, with the effect varying greatly by subject area.
Today agencies and tools use GEO as an umbrella term for everything to do with presence in AI answers. AEO (Answer Engine Optimization) and LLMO (Large Language Model Optimization) are also used for the same field. Google mentions AEO and GEO in its guide to generative AI features and classes both as SEO.
What do SEO and GEO have in common?
An AI answer with sources is created in two steps. The system looks for matching pages, either in its own index or by fetching them at the moment of the question. Then the language model writes an answer from these pages. Everything you know from technical SEO hinges on the first step.
- Reachability. The bot gets status 200 and is turned away neither by the robots.txt nor by bot protection. For ChatGPT this is OAI-SearchBot. According to OpenAI, ChatGPT does not show pages that block it in its search answers.
- Indexability. According to Google, a page must be indexed and eligible to appear in search with a snippet to be used in AI Overviews and AI Mode.
- Readability without JavaScript. Many AI systems only evaluate what is visible in the delivered HTML when they fetch a page directly. Our own test further down shows this too.
Google describes the situation clearly. Its AI features rely on the same ranking and quality systems as regular search. According to Google, you need no special markup, no llms.txt and no chopping of texts into small chunks for this.
Where do GEO and SEO differ?
| SEO | GEO | |
|---|---|---|
| Goal | Position in the search results and a click to your own page | Brand mention, recommendation and a link as a source in the answer |
| Input | Search query of a few words | Question in a conversation, often with context on budget, location or requirements |
| Result | List of links with snippets | Written answer with a handful of sources |
| Who fetches | Googlebot, Bingbot | Additionally the AI providers' bots, such as OAI-SearchBot and ChatGPT-User |
| Measurement | Search Console, rank tracking, web analytics | Prompt monitoring, cited sources, AI bots in log files |
| Data | Clicks and impressions per page | Traffic from AI Overviews sits in Search Console under the “Web” search type, without a separate breakdown |
Why are rankings alone not enough?
An AI Overview can answer above a page in position 1, and ChatGPT may recommend a competitor without anything changing in your rankings. Google also describes how AI Mode and AI Overviews fan a question out into several related searches. As a result, pages appear as sources that do not even rank for the original query. So you only see whether and how you appear in answers if you query the answers yourself.
What do AI systems see of your page?
On 30 October 2025 we built a test page with eight variants of a fictitious product. Each variant had a different price, and each price was placed somewhere else, for example visible in the HTML, loaded via JavaScript, only in JSON-LD or in hidden microdata. Then we asked ChatGPT, Gemini, Claude, Perplexity and Google AI Mode about the prices on the page.
| System | Prices found | Notable |
|---|---|---|
| ChatGPT | 3 of 8 | Only found the visible prices in the HTML |
| Gemini | 4 of 8 | Also found the price loaded via JavaScript |
| Claude | 0 of 8 | Recognized the products, but gave no price |
| Perplexity | 1 of 8, only after indexing | No access to the new page before that |
| Google AI Mode | 2 of 8, only after indexing | No answer before that |
- No system found a price that was only in the JSON-LD when fetching the page directly.
- Visible text in the delivered HTML was the most reliable source.
- Systems that rely on their own index did not know the new page at all at first.
The test covers a single page on one reference date. It shows a tendency and is no substitute for measuring your own website.
Structured data still makes sense. Google continues to recommend it for rich results, and the test could not measure what role it plays in indexing for AI answers.
Measuring GEO
- Create prompts the way your customers would ask, derived from your most important keywords.
- Query them regularly in several systems and record who is mentioned, recommended and linked as a source.
- Analyze in your log files which AI bots fetch which pages and which status they get.
- Check whether robots.txt and bot protection let the AI search bots through.
Common mistakes
- Bot protection blocks OAI-SearchBot or other AI providers' search bots without anyone noticing.
- The website blocks all OpenAI bots to rule out training and thereby also disappears from ChatGPT search. A separate group for GPTBot is enough to cover training.
- Prices, availability or key information only exist in JavaScript or in markup.
- A single prompt test is taken as the result. Answers fluctuate, so only series over several weeks are meaningful.
- The GEO project runs separately from the SEO team, even though both work on the same pages.
Frequently asked questions
Is GEO the new SEO?
GEO complements SEO and requires it. Google says that optimizing for its AI search is still SEO. What is new is mainly the goal, being mentioned in answers, and the way it is measured.
What is the difference between GEO, AEO and LLMO?
In practice the three terms mean the same field. GEO comes from the 2023 paper by Aggarwal and colleagues, AEO stands for Answer Engine Optimization, LLMO for Large Language Model Optimization.
How credible is GEO?
The basics are solid: reachability, indexing and clear content. Be careful with promises that sell certain phrasings or special files as levers. Google advises checking such advice against its own documentation.
Do I need dedicated GEO tools?
For the technical side, the tools from SEO are enough. They do not show mentions and sources in AI answers. For that you need prompt monitoring that queries the answers regularly.
Does schema markup help with ChatGPT and co.?
In our test, no system used prices that were only in the JSON-LD when fetching the page directly. Google says that its AI search does not require structured data. They are still worth it for rich results.
Is SEO still relevant?
Yes. AI Overviews and AI Mode draw on the Google index, and ChatGPT also needs pages that its search bot is allowed to fetch. Without a technically sound website there is no AI visibility.
Sources
Retrieved on October 15, 2026. If something is out of date or missing, let us know.
- Aggarwal et al., GEO: Generative Engine Optimization (arXiv 2311.09735, v1 Nov 16, 2023, v3 Jun 28, 2024, KDD 2024): https://arxiv.org/abs/2311.09735
- Google Search Central, Guide to generative AI features: https://developers.google.com/search/docs/fundamentals/ai-optimization-guide · As of Jul 10, 2026
- Google Search Central, AI features and your website: https://developers.google.com/search/docs/appearance/ai-features · As of Dec 10, 2025
- OpenAI, Crawlers (OAI-SearchBot, GPTBot, ChatGPT-User): https://developers.openai.com/api/docs/bots