Generative engine optimization (GEO): what it is and how to do it

What generative engine optimization means

A generative engine is any system that writes an answer instead of returning a list of links. ChatGPT search, Perplexity, Gemini, Microsoft Copilot and Google's AI Overviews and AI Mode all work this way. They retrieve pages, read them, and compose a response that may name your brand, quote your page or link to it.

GEO is the work of improving your odds of being in that response. It covers three layers:

• Access. AI crawlers can reach your pages, and the pages render as readable text.
• Understanding. The model can tell who you are, what you sell and what each page is about. That means clear entity data, schema and consistent facts across the web.
• Selection. When the model picks sources, your page is the one with the clearest, most specific, most quotable answer.

The goal is not traffic alone. Much of the value is being named as the answer when a buyer asks an AI assistant who to hire or what to buy, whether or not they click.

Where the term GEO came from

The name was coined in the paper "GEO: Generative Engine Optimization" by Pranjal Aggarwal, Vishvak Murahari, Tanmay Rajpurohit, Ashwin Kalyan, Karthik Narasimhan and Ameet Deshpande, with authors from Princeton University and IIT Delhi. It was first posted to arXiv in November 2023 and accepted to KDD 2024.

The researchers built GEO-bench, a benchmark of 10,000 queries drawn from nine datasets, and tested nine ways of rewriting a source page to see which ones made a generative engine use it more. Their main findings:

The best methods improved visibility by up to 40% over the baseline. Two details matter for small sites. First, the Cite Sources method raised visibility by 115.1% for pages ranked fifth in the underlying search results, while the top-ranked page's visibility fell on average. GEO can help a smaller site win citations from a bigger one. Second, the authors found results varied by domain, so what works for a recipe site may not work for a law firm.

The team also tested on Perplexity, a live engine, where Quotation Addition gave a 22% improvement on their word-count visibility metric.

GEO vs SEO

GEO does not replace SEO. Most AI engines retrieve pages through a search index, so if you do not rank or are not indexed, you are rarely in the pool to be cited. What changes is the final step.

If you want the long version, read GEO vs SEO. And if you are confused by the alphabet soup, SEO vs AEO vs GEO lays out how the three terms overlap.

What Google says about optimizing for AI answers

Google is the one AI search provider that has published detailed guidance, and it is worth reading because it pushes back on some GEO folklore.

Google's Search Central documentation on AI features says there are no additional requirements to appear in AI Overviews or AI Mode. Pages need to be indexed and eligible to show a snippet. You do not need new machine-readable files, AI-specific text files or special schema to appear.

Its May 2025 post for site owners lists what does matter: unique content that is not commodity, a good page experience, crawlable pages, preview controls set on purpose, structured data that matches visible content, good images and video, and measuring conversions as well as clicks.

So for Google, GEO is mostly very good SEO plus content written to be quoted. Other engines run their own crawlers and indexes, which is where the technical side of GEO earns its keep.

Generative engine optimization strategies that work

Here is the practical playbook, ordered roughly by effort.

1. Let AI crawlers in. Each engine has its own bot. OpenAI's documentation says OAI-SearchBot controls whether a site appears in ChatGPT search answers, while GPTBot is for model training and can be blocked separately. Perplexity says PerplexityBot is what surfaces sites in its results. Check your robots.txt and your CDN or firewall rules, which often block these bots without anyone noticing.

2. Make pages readable without JavaScript. Many AI crawlers fetch raw HTML. If your content only appears after client-side rendering, some engines see an empty page.

3. Answer first. Put a direct two to four sentence answer right under each question-style heading. That passage is what gets lifted.

4. Add statistics, quotations and sources. This is the core finding of the GEO paper. Name the source in the copy and link to it. A specific number with an origin beats a vague claim.

5. Use structure the model can parse. Short paragraphs, descriptive headings, lists, comparison tables and a FAQ section.

6. Get your entity data straight. Organization schema, consistent name, address and description across your site, directories, social profiles and review sites. Models build their picture of you from all of it.

7. Earn mentions elsewhere. AI answers lean on third-party sources: reviews, comparison articles, forums and industry lists. Being mentioned in those places affects whether you are recommended.

8. Publish an llms.txt file. llms.txt is a proposal from Jeremy Howard, published in September 2024, for a Markdown file at your site root that gives AI agents a curated map of your key pages. Google says it does not need one, so treat it as a cheap, low-risk extra rather than a ranking factor. Our free llms.txt generator builds one in a minute.

9. Keep content current. Out-of-date pricing or facts on your site give engines a reason to cite a competitor instead.

How to measure GEO

There is no Search Console for ChatGPT. Measuring GEO means tracking prompts.

1. Build a prompt list. 20 to 50 questions your buyers actually ask: "best [service] in [city]", "[your brand] vs [competitor]", "how do I [problem you solve]".
2. Run them on a schedule across ChatGPT, Perplexity, Gemini and Google AI Overviews.
3. Record three things: whether you are mentioned, whether you are cited with a link, and which competitors appear.
4. Watch the trend, not single runs. AI answers vary from one run to the next, so one check proves little.

Referral traffic in analytics from chatgpt.com, perplexity.ai and similar domains is a useful second signal, though it undercounts because many people never click.

You can run this by hand in a spreadsheet, use one of the GEO tools on the market, or have someone run it for you. To see where you stand today, start with the free AI visibility checker.

GEO for different kinds of business

The GEO paper's authors found that what works varies by domain. In practice the priorities shift with your business type.

Whatever the type, the order of work is the same: access first, then entity clarity, then content, then reputation. There is no point polishing copy on a page an AI crawler cannot reach.

Common GEO mistakes

• Blocking AI search bots by accident while trying to block training bots. They are separate user agents. Block GPTBot if you want, but keep OAI-SearchBot.
• Keyword stuffing for AI. The GEO paper found it does little or nothing.
• Chasing hacks over fundamentals. If you are not indexed or your content is thin, no AI trick fixes that.
• Measuring with one prompt run. Answers vary. Track a set of prompts over time.
• Writing for the model instead of the reader. Google's guidance is explicit that people-first, original content is what it wants to surface.

Doing GEO yourself vs hiring help

A small site can do most of this in-house: fix robots.txt, add schema, rewrite key pages answer-first, and track 20 prompts in a spreadsheet each month. The hard parts are keeping it going and knowing which fixes matter most.

If you would rather hand it off, our GEO services run on a flat $299 a month. Month one fixes what the free audit finds (AI crawler access, schema, llms.txt, entity data and content gaps), then we keep optimizing and monitor up to 50 prompts a month across ChatGPT, Perplexity, Gemini and Google AI Overviews. See how our generative engine optimization services work, or run the free audit first.