What Is Generative Engine Optimization? GEO vs AEO vs SEO

Generative engine optimization (GEO) is the practice of shaping your content and brand presence so AI engines such as ChatGPT, Gemini, Perplexity and Claude use your pages, and name your brand, when they write an answer. It sits on top of SEO and extends answer engine optimization (AEO), which is about making a single page easy to quote.

The three terms overlap more than the acronyms suggest, and most guides blur them. This one defines each, compares them in tables, walks through what the original research measured (and what it did not), shows how AI crawlers decide what they can read, and ends with a checklist and a measurement method you can run this week. Every number links to the source it came from, checked in November 2026.

Key takeaways

• GEO was named in a research paper, "GEO: Generative Engine Optimization", first posted on arXiv in November 2023 and published at KDD 2024. It is a research term for a real shift: engines now write one answer from several sources.
• In that paper, adding citations, quotations and statistics improved visibility, while keyword stuffing, the classic SEO habit, did not perform well.
• SEO is still the base layer. Google's documentation says a page must be indexed and eligible to show with a snippet to appear in its AI features.
• Each AI engine has its own crawlers with different rules. Blocking the wrong user agent can remove you from an engine's search answers or do nothing at all.
• GEO is measured in answers, not rankings: run a fixed prompt set on several engines and record whether your brand is named and which of your URLs are cited.
• Start by scoring your pages with the free GEO readiness scorecard, then fix crawler access before you rewrite content.

What is generative engine optimization?

Generative engine optimization is the work of making a brand's content more likely to be included, quoted and cited in answers written by generative AI engines. A generative engine retrieves sources, then writes one answer from them. The question stops being "where do I rank?" and becomes "am I in the answer, and am I named as the source?"

The term comes from a paper by Pranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan and Ameet Deshpande, "GEO: Generative Engine Optimization" (arXiv 2311.09735, Cornell University, version 3 dated June 2024). The authors describe GEO as a framework for helping content creators improve how visible their content is in generative engine responses. They built GEO-bench, a benchmark of 10K queries from multiple sources, to test it.

Aggarwal et al. report that GEO "can boost visibility by up to 40% in generative engine responses", measured on their own benchmark, and that the effect of each strategy varies by domain.

That 40% is a research result on the paper's own metrics, not a traffic forecast. It remains the most cited primary evidence that page-level changes affect whether an AI engine uses a source.

How is GEO different from AEO and SEO?

SEO optimises for ranking in a list of results, AEO optimises a passage so it can be lifted as an answer, and GEO optimises a brand's whole footprint so engines trust and cite it. The table sets the three side by side.

Many practitioners use AEO and GEO interchangeably, and that is fine in conversation. The useful distinction is scope: AEO is how you write one page; GEO is everything that decides whether engines trust your brand. For the wider vocabulary, see the AI search glossary for marketers.

Does GEO replace SEO?

No. Google's documentation on AI features and your website says that to appear as a supporting link in AI Overviews or AI Mode, a page must be indexed and eligible to be shown in Google Search with a snippet. It adds that no other special optimizations are required. For Google's AI answers, good SEO is the entry ticket.

The same page says you do not need to create new machine-readable files, AI text files or markup to appear in these features. That matters when vendors sell shortcuts. Crawlable, indexed, useful pages come first; everything else is a refinement.

What did the GEO research actually find?

The paper tested nine ways of rewriting a source page, including Authoritative, Statistics Addition, Keyword Stuffing, Cite Sources, Quotation Addition, Easy-to-Understand, Fluency Optimization, Unique Words and Technical Terms. It scored visibility with two metrics: Position-Adjusted Word Count, which rewards a source whose words appear more, and earlier, in the answer, and Subjective Impression, a language-model judgement of how relevant and influential the cited material is.

All figures are from the full HTML version of the paper on arXiv. Read them as relative changes on research metrics in a controlled setting. They show direction (evidence and quotations help, stuffing keywords does not), not how much traffic any site will gain. They also predate today's engines, so treat them as a hypothesis to test on your own topic. Our guide to GEO experiment design shows how.

Worked example: one sentence, rewritten the GEO way

Here is the same claim written two ways. The second applies the paper's top methods (cite a source, add a figure) using only facts from the source.

• Before: "Optimising for AI engines can make a big difference to visibility."
• After: "In a paper published at KDD 2024, Aggarwal et al. report that generative engine optimization can boost visibility by up to 40% in generative engine responses, on their GEO-bench benchmark of 10K queries."

The second sentence names who found it, where it was published, how big the test was and how large the effect was, so an engine can quote it and attribute it. The rule that keeps this safe: every added number must come from a source you can link. An invented statistic is fabrication, and it is the fastest way to lose trust with readers and engines.

What does a quotable passage look like?

A quotable passage is complete without the text around it. It names the entity in full, states one claim, and carries its source. AI engines lift passages, not pages, so each section should survive being copied alone.

The pattern is explained further in our guide to the sentence structure that gets quoted in AI answers, and the page-level layout is covered in how to build an FAQ page for AI search.

How do AI crawlers decide what they can read?

Before an engine can cite a page, its crawler must be allowed to fetch it. Each vendor documents its own user agents, and search, training and user-triggered fetches are separate. This is the part of GEO most often skipped, and it is cheap to check.

Sources: OpenAI, overview of crawlers, Perplexity, crawlers documentation and Google Search Central, checked November 2026. Vendors change these pages, so confirm before you edit robots.txt. Our AI crawler user agents list explains what to allow and what to block.

How to do generative engine optimization: a checklist

1. Open the door. Check robots.txt, your CDN and your firewall for Googlebot, OAI-SearchBot and PerplexityBot. A blocked crawler means no citation, whatever the content.
2. Get the SEO basics right. Indexed, snippet-eligible pages with clear titles and internal links are the base for Google's AI features.
3. Answer first. Put a direct, complete answer in the first one or two sentences under each question-style heading, then add depth.
4. Cite, quote and count. Link primary sources, quote named documents or experts, and add figures you can source. These were the strongest methods in the GEO paper.
5. Describe your brand one way everywhere. Use the same one-sentence description on your site, in your schema and on your profiles. See how to write a brand description for AI search.
6. Mark up what is on the page. Article and FAQ schema should match the visible text; see schema markup for AI search engines.
7. Earn mentions beyond your site. Engines retrieve other sites as well as yours, so reviews, community threads and press that name you can shape the answer.
8. Measure with a fixed prompt set. Ask the same 10 to 20 buyer questions on a schedule and log whether you are named and which URLs are cited.

For a page-by-page pass, use the AEO audit checklist. For why some sites get trusted more than others, read why AI engines trust some websites more.

Where should you start: SEO, AEO, GEO or measurement?

Start with the first row that matches your situation. Fixing an earlier layer is usually cheaper than polishing a later one.

How do you measure GEO?

GEO is measured in answers, not positions. The core metric is the share of a fixed prompt set in which an engine names your brand, alongside which of your URLs it cites and which competitors it names instead. Answers change between runs and between engines, so you need repeated runs across several engines rather than a single snapshot.

A measurement method you can run by hand

1. Write 10 to 20 prompts a real buyer would ask, mixing category questions, comparison questions and "best for" questions. Our guide to which prompts to track helps you choose.
2. Run each prompt in each engine you care about, in a fresh session, on the same day each week.
3. Record three things per run: whether your brand is named, which URLs are cited, and which competitors appear.
4. Divide prompts with a mention by prompts run. As arithmetic only: if a brand were named in 6 of 20 prompts, its mention share would be 30%. That figure is an illustration of the formula, not a benchmark.
5. Change one thing at a time, then compare the next four weekly runs against the baseline.

Personal history and location can shift answers, so read whether personalization changes AI answers before you trust a single run. The longer method is in tracking an AI search visibility score week over week and how to measure AI search share of voice.

Noma by Nurdd is an AI search visibility platform that tracks whether a brand appears in answers from 13 AI engines, including ChatGPT, Gemini, Perplexity and Claude, polled daily. It shows citation frequency by category, how competitors compare and which recommendations to act on, so the weekly spreadsheet becomes a dashboard you do not have to maintain.

Want to see where AI engines name your brand today? Book a demo and the Nurdd team will show you your own category prompts, which competitors appear in your place and which pages to fix first.

What are the most common GEO mistakes?

• Treating GEO as a replacement for SEO. Google says pages need to be indexed and snippet-eligible to appear in its AI features, which is SEO work.
• Keyword stuffing. The GEO paper describes it as a method that does not perform well on generative engines.
• Adding numbers without sources. The paper's strongest methods add evidence. An unsourced figure is not evidence.
• Blocking the wrong bot. Blocking GPTBot concerns training; blocking OAI-SearchBot removes you from ChatGPT search answers. Know which one you are changing.
• Measuring rankings only. A page can rank and never be cited, or be cited without ranking. Track both.
• Judging from one run. One answer in one engine is an anecdote. Repeat the prompts.

Does llms.txt help with GEO?

llms.txt is a community proposal to publish a plain-language summary of a site for AI agents. It is not a standard every engine has adopted. For Google specifically, its AI features guidance says you do not need new AI text files to appear in them.

Treat llms.txt as a low-cost extra for agents that read it, never as a substitute for crawlable, well-structured pages. A consistent entity description matters more; see Wikidata and sameAs for brands for how engines learn who you are.

Which engine should you optimise for first?

Optimise for the engines where your buyers ask questions, then check the others. Engines differ in how they retrieve sources, so a change that helps in one may not show in another. Our comparison of Perplexity, ChatGPT and Gemini for brands covers the differences, and there are separate guides on getting ChatGPT to recommend your brand and getting traffic from Perplexity.

Should you do GEO in-house or hire help?

Do it in-house if you have someone who can edit pages, fix crawler access and run a weekly prompt set. Hire help if the work spans content, technical access and third-party mentions and nobody owns it. The trade-offs are laid out in AEO agency vs in-house tooling. If you want it run for you, Nurdd offers generative engine optimization services, measured in Noma.

How we researched this

This guide was written by Simran Rawat. We read the GEO paper on arXiv (abstract and full HTML version), Google Search Central's AI features documentation, OpenAI's and Perplexity's crawler documentation, and several top-ranking articles on GEO from Search Engine Land, Frase, Mangools and Optimizely, all checked in November 2026. Every number comes from the primary sources and is stated as they give it. We did not run our own experiments for this post and cite no Nurdd performance figures.

See your brand in AI answers

The quickest way to know where you stand is to see your own prompts run. Book a demo and you will see which of 13 AI engines name your brand, which competitors appear where you do not, and which cited pages to fix first. Noma offers a 3-day free trial at nomahq.app; a card is taken at signup and billing starts on day 3, with no charge if you cancel before then.

Frequently asked questions

What is generative engine optimization in simple terms?

Generative engine optimization is making your content and brand easy for AI engines like ChatGPT, Gemini, Perplexity and Claude to use and cite when they write an answer. It focuses on whether your brand is named and your pages are cited, rather than where a page ranks.

What is the difference between GEO and AEO?

AEO is page-level: structuring a page so an engine can extract a direct answer. GEO is broader: everything that makes AI engines trust and cite your brand across a topic, including sourced evidence, a consistent brand description, third-party mentions and crawler access. Many people use the terms interchangeably.

Is GEO replacing SEO?

No. Google says a page must be indexed and eligible to show with a snippet to appear as a supporting link in AI Overviews or AI Mode, and that no special optimization is needed beyond that. SEO remains the foundation GEO builds on.

Who coined the term generative engine optimization?

The term comes from the paper "GEO: Generative Engine Optimization" by Pranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan and Ameet Deshpande, first posted on arXiv in November 2023 and published at KDD 2024.

What GEO techniques work best?

In the GEO paper, Statistics Addition and Quotation Addition showed strong improvements, with the best methods improving on the baseline by 41% on Position-Adjusted Word Count and 28% on Subjective Impression. Keyword stuffing did not perform well. Results varied by domain, so test changes on your own topic.

Do I need an llms.txt file for GEO?

It is optional. llms.txt is a proposal for summarising a site for AI agents, not a universal standard, and Google says you do not need new AI text files to appear in its AI features. Crawlable, clearly structured pages matter far more.

How do I know if AI engines cite my website?

Ask a fixed set of buyer questions in each engine on a schedule and record whether your brand is named and which URLs are cited. Check your server logs for the engines' crawlers too. A tracker such as Noma can automate the runs across 13 engines.