How to rank in ChatGPT: a GEO and AEO playbook

To rank in ChatGPT you need three things: a page its crawler is allowed to fetch, an answer a model can lift in one clean chunk, and sources it can trust enough to cite. AI answer engines do not show ten blue links. They read a handful of pages and name a few of them, so the goal is to become one of the pages they quote.

Optimizing to be cited by generative engines has a name, Generative Engine Optimization, and the eight steps below turn it into concrete changes you can ship this week.

AI search is a different surface

ChatGPT Search, Perplexity, Gemini, and Google AI Overviews all answer a question directly and cite the pages behind the answer. The user reads the synthesized reply and clicks through only sometimes. Your target moves from ranking a link to earning a citation inside the generated answer, and the page that gets cited is usually the one that answered the question most clearly and backed it with evidence.

GEO, AEO, and LLM SEO, defined

Generative Engine Optimization (GEO) is shaping content so generative engines cite it. The term comes from research by Aggarwal and colleagues, presented at KDD 2024, which measured what actually changes whether a page is used in an AI answer.

Answer Engine Optimization (AEO) is the narrower goal of becoming the direct answer to a specific question. LLM SEO is the umbrella name people use for the same work across large language model products. The three labels describe one practice: make your page the easiest correct answer for a model to reuse.

The playbook: eight steps to get cited

1. Let the AI crawlers in. Add explicit rules in robots.txt for GPTBot and OAI-SearchBot from OpenAI, PerplexityBot, ClaudeBot, and Google-Extended. If a crawler cannot fetch the page, none of the steps below can help.
2. Answer the question in the first two sentences. Models lift self-contained passages, so put the direct answer at the top and expand underneath it, rather than opening with background.
3. Give every section a question and a short answer. Use plain headings that match how people ask, and keep one idea per paragraph so a passage reads correctly on its own, outside the page.
4. Add machine-readable structure. Mark up the page with Article, FAQPage, and HowTo schema so engines can map questions to answers without inferring the structure from raw text.
5. Cite sources and use concrete numbers. The GEO study found that adding cited sources, direct quotations, and statistics raised a page's visibility in AI answers by up to 40 percent. Vague prose loses to a sourced number.
6. Make the author and publisher clear. Name a real author with a short bio, keep the organization consistent across the site, and link to verifiable profiles so the model can attach trust to the source.
7. Keep the page fresh. Show a visible last-updated date and set dateModified in the markup. Models have knowledge cutoffs and live-retrieval engines prefer recent pages, so a current date helps you get picked on fast-moving topics.
8. Publish an llms.txt and earn mentions elsewhere. A short llms.txt gives models a clean summary of your site, and independent pages that mention you raise the odds that an engine cites you rather than a competitor.

How to check whether it is working

Ask the engines directly. Put your target question to ChatGPT Search, Perplexity, and Google, and see whether your page is named. Watch analytics for referral traffic from chatgpt.com and perplexity.ai, and track the AI performance view in Bing Webmaster Tools, which reports how often your content surfaces in AI answers. Treat each cited answer as a ranking you earned and each missed one as a page to tighten.

Frequently asked questions