AI Search Visibility
What Is Generative Engine Optimization? GEO vs AEO vs SEO
GEO, AEO and SEO defined in one sentence each, compared in a table, with the original GEO research findings and a checklist you can use this week.
Simran Rawat1 Nov 202613 min read
Generative engine optimization (GEO) is shaping content so AI engines such as ChatGPT, Gemini, Perplexity and Claude use it and name your brand in their answers. AEO structures a single page so an engine can lift a direct answer, and SEO helps search engines understand and rank your pages. In the GEO paper by Aggarwal et al., citing sources, adding quotations and adding statistics worked best, while keyword stuffing offered little to no improvement.
Generative engine optimization (GEO) is the practice of shaping your content so that AI engines such as ChatGPT, Gemini, Perplexity and Claude use it, and name your brand, when they write an answer. Answer engine optimization (AEO) is the practice of structuring a page so that an engine can lift a direct answer from it. Search engine optimization (SEO) is, in Google's words, about helping search engines understand your content and helping people find your site through search.
The three overlap more than the acronyms suggest. SEO gets a page crawled, indexed and ranked; AEO makes each page easy to quote; GEO is the wider effort to be the source an AI engine trusts and cites across a whole topic. This guide defines each one, shows where they differ in a comparison table, summarises what the original GEO research actually measured, and gives a checklist you can apply this week. Every number below links to the primary source it comes from, checked on 1 November 2026.
Key takeaways
- GEO was named in a research paper, "GEO: Generative Engine Optimization" by Pranjal Aggarwal and co-authors, first posted on arXiv in November 2023 and accepted to KDD 2024.
- In that paper, adding citations, quotations and statistics to a page were the top methods, with a 30–40% relative improvement on the paper's visibility metric. Keyword stuffing offered little to no improvement.
- Google says there are no extra requirements or special optimizations needed to appear in AI Overviews or AI Mode: a page has to be indexed and eligible to show with a snippet. SEO is still the base layer.
- Each AI engine has its own crawlers. Blocking OAI-SearchBot keeps a site out of ChatGPT search answers, and blocking PerplexityBot keeps it out of Perplexity's search results.
- GEO is measured by asking engines a fixed set of prompts and recording whether your brand is named and your pages are cited, not by rankings.
- Start with the free GEO readiness scorecard to see how your own pages score.
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, so the question is no longer "where do I rank?" but "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, version 3, June 2024). The authors describe GEO as a framework that helps content creators improve how visible their content is in generative engine responses, and they introduced GEO-bench, a benchmark of 10K queries drawn 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 metrics, 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 is still the most cited primary evidence that page-level changes affect whether an AI engine uses a source.
What is answer engine optimization?
Answer engine optimization is the page-level discipline of writing and structuring content so an engine can extract a complete, accurate answer from it. In practice that means a question-style heading, a direct answer in the first sentence or two beneath it, short paragraphs, lists and tables for comparisons, and an FAQ that matches what people actually ask. Our guide to building an FAQ page for AI search shows the pattern.
Many practitioners use AEO and GEO interchangeably. The useful distinction is scope: AEO is how you write one page so it can be quoted; GEO is everything that decides whether engines trust and cite your brand across a topic, including mentions on other sites, a consistent brand description and access for AI crawlers.
What is search engine optimization?
Search engine optimization is, according to Google's SEO Starter Guide (last updated December 2025), about helping search engines understand your content and helping users find your site and decide whether to visit it through a search engine. The levers are familiar: crawlable pages, useful content, descriptive titles, internal links and a good page experience.
SEO has not been replaced. Google's own guidance on AI features and your website (last updated December 2025) 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, and that no other special optimizations are necessary. For Google's AI answers, good SEO is the entry ticket.
GEO vs AEO vs SEO: what is the difference?
| SEO | AEO | GEO | |
|---|---|---|---|
| Goal | Rank pages in search results and earn the click | Have a page's answer extracted and shown | Be used, quoted and named in AI-written answers across a topic |
| Where the result appears | Search results pages | Featured answers, AI Overviews, voice and chat answers | ChatGPT, Gemini, Perplexity, Claude, Google AI Mode and other engines |
| Unit of work | The page and the site | The passage under each heading | The brand, its pages and what other sites say about it |
| Main levers | Crawlability, relevance, links, page experience | Question headings, direct answers, lists, tables, FAQ | Cited sources, quotations and statistics, consistent entity description, third-party mentions, AI crawler access |
| What hurts | Thin or duplicate pages, blocked resources | Answers buried deep in the page | Keyword stuffing (little to no gain in the GEO paper), unsourced claims, blocked AI crawlers |
| How you measure | Rankings, impressions, clicks | Whether the snippet or answer uses your text | Share of prompts where your brand is named, and which of your URLs are cited |
What did the GEO research actually find?
The paper tested nine ways of rewriting a source page: Authoritative, Statistics Addition, Keyword Stuffing, Cite Sources, Quotation Addition, Easy-to-Understand, Fluency Optimization, Unique Words and Technical Terms. It measured 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 score judged by a language model that covers how relevant, influential and prominent the cited material is.
| Finding | What the paper reports |
|---|---|
| Top methods | Cite Sources, Quotation Addition and Statistics Addition: a relative improvement of 30–40% on Position-Adjusted Word Count and 15–30% on Subjective Impression. |
| Best methods overall | Improved on the baseline by 41% on Position-Adjusted Word Count and 28% on Subjective Impression. |
| Keyword stuffing | Widely used for SEO, but offered little to no improvement on generative engine responses. |
| Real engine test | On Perplexity.ai, Quotation Addition performed best on Position-Adjusted Word Count, a 22% improvement over the baseline; the paper reports visibility improvements of up to 37% there. |
| Domains differ | The effectiveness of each strategy varies across domains, so test on your own topic. |
All figures are from the 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 a given site will gain.
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 and add a statistic) using only facts from that source.
- Before: "Optimising for AI engines can make a big difference to visibility."
- After: "In a paper accepted to KDD 2024, Aggarwal et al. report that generative engine optimization can boost visibility by up to 40% in generative engine responses, with citations, quotations and statistics working best."
The second sentence names who found it, where, and how much, so an engine can quote it and attribute it. The rule that makes this safe is simple: every added number must come from a source you can link. Adding invented statistics would be fabrication, and it is the fastest way to lose trust with readers and engines alike.
How do AI engines find and choose sources?
Before an engine can cite a page, its crawler has to be allowed to fetch it. Each vendor documents its own user agents, and the rules differ between search, training and live user fetches.
| User agent | Vendor | What it does, per the vendor's documentation |
|---|---|---|
| OAI-SearchBot | OpenAI | Surfaces websites in ChatGPT's search features. Sites that opt out are not shown in ChatGPT search answers. |
| GPTBot | OpenAI | Crawls content that may be used to train OpenAI's foundation models. Disallowing it signals that content should not be used for training; search inclusion is controlled by OAI-SearchBot. |
| ChatGPT-User | OpenAI | Handles certain user-initiated actions; not used for automatic crawling or to decide search inclusion. |
| PerplexityBot | Perplexity | Surfaces and links websites in Perplexity's search results; not used to train foundation models. |
| Perplexity-User | Perplexity | Visits a page when a user's question needs it; generally ignores robots.txt because a user requested the fetch. |
| Googlebot | Indexes pages for Google Search, which AI Overviews and AI Mode draw on; a page must be indexed and snippet-eligible. |
Sources: OpenAI crawler overview, Perplexity crawlers and Google Search Central, all checked 1 November 2026. Once a page can be fetched, what gets cited depends on how directly a passage answers the question and how much the engine trusts the source; see why AI engines trust some websites more.
Does llms.txt help with GEO?
llms.txt is a proposal by Jeremy Howard, published on 3 September 2024, to standardise an /llms.txt file that gives AI agents a plain summary of a website. It is a proposal, not a standard that every engine has adopted. Google is explicit for its own features: its AI features guidance says you don't need to create new machine-readable files, AI text files or markup to appear in them.
Treat llms.txt as a low-cost extra for agents and tools that read it, never as a substitute for crawlable, well-structured pages. Nurdd publishes one, alongside a canonical facts page for AI engines, so that the same brand description appears everywhere.
How to do generative engine optimization: a checklist
- 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.
- Get the SEO basics right. Indexed, snippet-eligible pages with clear titles and internal links are the base for Google's AI features.
- Answer first. Put a direct, complete answer in the first one or two sentences under each question-style heading, then add depth.
- Cite, quote and count. Link primary sources, quote named experts or documents, and add statistics you can source. These were the top methods in the GEO paper.
- Describe your brand one way everywhere. Use the same one-sentence description on your site, your schema and your profiles; our guide to writing a brand description for AI search walks through it.
- Mark up what is on the page. Article and FAQ schema should match the visible text; see schema markup for AI search engines.
- 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.
- Measure with a fixed prompt set. Ask the same 10 to 20 buyer questions every week and log whether you are named and which URLs are cited.
For a page-by-page pass, use the AEO audit checklist.
What are the most common GEO mistakes?
- "GEO replaces SEO." It doesn't. Google says pages need to be indexed and snippet-eligible to appear in its AI features, which is SEO work.
- Keyword stuffing. In the GEO paper it offered little to no improvement in generative engine responses.
- Adding numbers without sources. The paper's statistics method works by adding evidence. An unsourced or invented figure is not evidence.
- Blocking the wrong bot. Blocking GPTBot opts out of 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; see how to measure AI search share of voice.
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. Because answers change between runs and between engines, you need repeated runs across several engines rather than a single snapshot. Our guides on tracking an AI search visibility score and on how ChatGPT, Gemini and Perplexity differ cover the method.
You can run that by hand in a spreadsheet, or automate it. Noma, Nurdd's AI search visibility platform, tracks where 13 AI engines, including ChatGPT, Gemini, Perplexity and Claude, name your brand every day, and shows which competitors are named instead and which pages the answers draw on. If you want to compare tools first, read our guide to the best AI search visibility platforms.
Want to see where AI engines name your brand today? Book a demo and the Nurdd team will run your own buyer prompts and show you the results.
Where to start: GEO, AEO or SEO first?
| If this is true | Start with | First step |
|---|---|---|
| Key pages are not indexed or have technical errors | SEO | Fix crawling and indexing, then internal links |
| Pages rank but answers are buried | AEO | Rewrite intros and add question headings with direct answers |
| Pages are quotable but engines name competitors | GEO | Add sourced evidence, align your brand description and earn third-party mentions |
| You don't know whether engines name you | Measurement | Run a fixed prompt set across several engines |
For a deeper look at getting named in a specific engine, read how to get ChatGPT to recommend your brand and how to get traffic from Perplexity. If you are deciding who should do the work, see AEO agency vs in-house.
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 SEO Starter Guide and its AI features page, OpenAI's and Perplexity's crawler documentation, and the llms.txt proposal, all checked on 1 November 2026. Every number in this post comes from those primary sources and is quoted as they state it. We also reviewed the top-ranking articles on GEO, AEO and SEO to cover the questions they answer. We did not run our own experiments for this post and cite no Nurdd performance figures.
Want this done for you? Nurdd runs it as a managed service: see our generative engine optimization services.
Sources
- Aggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan and Deshpande, "GEO: Generative Engine Optimization", arXiv 2311.09735 (v1 November 2023, v3 June 2024; accepted to KDD 2024), and full HTML version.
- Google Search Central, SEO Starter Guide, last updated December 2025.
- Google Search Central, AI features and your website, last updated December 2025.
- OpenAI, Overview of OpenAI crawlers, checked 1 November 2026.
- Perplexity, Perplexity crawlers, checked 1 November 2026.
- Jeremy Howard, The /llms.txt file, 3 September 2024.
- nomahq.app, for the 13 engines tracked, checked 1 November 2026.
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 the Nurdd team will show you which AI engines name your brand, which competitors appear where you do not, and which pages to fix first. You can also score your pages today with the GEO readiness scorecard.
Frequently asked questions
What is generative engine optimization in simple terms?
Generative engine optimization is making your content 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 that 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 accepted to KDD 2024.
What GEO techniques work best?
In the GEO paper, citing sources, adding quotations and adding statistics performed best, with a 30–40% relative improvement on its Position-Adjusted Word Count metric. Keyword stuffing offered little to no improvement. 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 2024 proposal for summarising a site for AI agents, not a universal standard, and Google says you don't 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 every week and record whether your brand is named and which URLs are cited. Check your server logs for the engines' crawlers too. An AI search visibility tracker can automate the runs across many engines.