What Is GEO? 7 Ways to Get Cited by ChatGPT and Perplexity
GEO means generative engine optimization: shaping a page so that AI search engines such as ChatGPT, Perplexity and Google AI Overviews can read it and cite it as a source. In this guide the Sag3o editorial team explains how GEO differs from SEO, how AI engines decide which sources to cite, and 7 ways to make your site a better candidate.
Key takeaways
- GEO does not replace SEO. The technical base is the same; the difference is content an AI can lift directly.
- ChatGPT, Claude and Perplexity each have a dedicated search crawler (OAI-SearchBot, Claude-SearchBot, PerplexityBot). Allow them in robots.txt.
- As of Vercel's late-2024 measurements, major AI crawlers do not execute JavaScript; keep your content in the raw HTML.
- Definitions, a top summary, question headings, structured data and an update date are the parts an AI can lift most easily. Nobody can guarantee a citation, but each can be checked.
How is GEO different from SEO?
SEO aims for a ranking and a click. GEO aims to be the source an AI quotes while writing an answer: the user may never visit your site, but sees your brand and wording. The technical base is the same; the difference is the content layer.
| Aspect | SEO | GEO |
|---|---|---|
| Goal | Rankings and clicks | Citations and brand mentions |
| Reader | Googlebot and people | AI search crawlers and language models |
| Content shape | Keywords, titles, internal links | Definitions, summaries, Q&A, tables |
| Crawler control | Allow Googlebot | Also allow OAI-SearchBot and other AI crawlers |
How do AI search engines decide which sources to cite?
First, they have to fetch the page. OpenAI uses OAI-SearchBot for search and GPTBot for training, set independently in robots.txt; a site that blocks OAI-SearchBot is not shown in ChatGPT search answers (it may still appear as a navigational link). Anthropic documents ClaudeBot, Claude-SearchBot and Claude-User; Perplexity runs PerplexityBot and Perplexity-User. Google AI Overviews uses Googlebot, and Google states that Google-Extended does not affect inclusion or ranking in Search.
Second, they have to read it. Vercel's crawler analysis from December 2024 found that no major AI crawler renders JavaScript: OpenAI's crawlers downloaded JS files in about 11.5% of requests and Claude in about 23.84%, but neither executed them. GPTBot, Claude, AppleBot and PerplexityBot together made nearly 1.3 billion fetches in a month, against roughly 4.5 billion for Googlebot. Content that appears only after client-side rendering is invisible. When we scanned our own home page with Sag3o, the raw-HTML-to-rendered ratio was 1.0; on a large government portal we found 0.72, meaning almost a third of its content only appears after JavaScript runs.
Third, they have to choose it. Definitions, answers under question headings and tables with column headers are the easiest parts to lift; the update date and outbound sources show whether the page is current and backed by evidence.
7 ways to get cited
- Open with a definition. State the topic in one "X is ..." sentence in the first paragraph, 20 to 70 words long.
- Put a summary up top. A bulleted list of takeaways right after the H1 gives the AI a ready-made answer.
- Write headings as questions. "How do I check my site?" matches a user's prompt better than "Site check".
- Add structured data. Article, FAQPage and BreadcrumbList markup tell machines who wrote it, when, and where the Q&A is.
- Allow the search crawlers. Permit OAI-SearchBot, Claude-SearchBot and PerplexityBot; the training bots GPTBot and ClaudeBot are a separate decision. An llms.txt file (a September 2024 proposal) can also list your main pages for AI tools.
- Ship content in the raw HTML. The body, tables and FAQ must be server-rendered, not loaded by JavaScript afterwards.
- Show an update date. Display it on the page, fill dateModified in your structured data, and keep the page current.
None of these seven guarantees a citation. Each removes a reason for an AI to skip you: it cannot read the page, or cannot tell what it claims. Items 1, 2 and 6 are usually the cheapest place to start.
How do you check your own site?
Every item above can be verified by a program: whether the first paragraph has a definition, how many headings are questions, which AI crawlers robots.txt allows, how much raw and rendered HTML differ, and how old the date is. Sag3o runs these checks and returns a report with the pass rule and evidence for each. Enter a URL to see your result.
Frequently asked questions
Do I need a separate site for GEO?
No. GEO and SEO run on the same site; what changes is how the content is written and which crawlers you allow. Adding a definition, a summary and a Q&A to existing articles usually beats building something new.
Should I block AI crawlers?
It depends. To be cited by ChatGPT or Perplexity, allow OAI-SearchBot, Claude-SearchBot and PerplexityBot. If you do not want your content used for training, block GPTBot and ClaudeBot; the official docs confirm the settings are independent.
How long does it take to see results?
There is no fixed timeline and nobody can promise an outcome. Use Sag3o first to confirm the technical side is in place, then search your own topic in ChatGPT and Perplexity now and then and see whether your site appears among the sources.
Sources
- OpenAI: Overview of OpenAI Crawlers (OAI-SearchBot, GPTBot, ChatGPT-User)
- Anthropic: Does Anthropic crawl data from the web (ClaudeBot, Claude-SearchBot, Claude-User)
- Perplexity: Perplexity Crawlers (PerplexityBot, Perplexity-User)
- Google Search Central: List of Google's common crawlers (Google-Extended and ranking)
- Vercel: The rise of the AI crawler (December 2024, AI crawlers do not render JavaScript)
- llmstxt.org: The /llms.txt file proposal (September 2024)
- Google: Generative AI in Search, AI Overviews rollout (May 2024)