For two decades, search engine optimization (SEO) meant ranking on a page of blue links. Now, many questions are answered directly by AI: search engines show AI-generated summaries, and people ask chat assistants with web search. When an AI writes the answer, the question becomes: will it cite you? This emerging discipline is often called generative engine optimization (GEO).

Where the term comes from

A 2023 research paper titled GEO: Generative Engine Optimization, by researchers from Princeton University and other institutions, studied how content changes affect visibility in AI-generated answers. In their experiments, adding citations, quotations from relevant sources and statistics to content tended to increase visibility in generative engine responses, while traditional keyword stuffing performed poorly. The results come from a specific experimental setup, but they point in a sensible direction.

How AI answers find sources

AI search systems typically:

  1. interpret the question;
  2. search an index (often a conventional web search index);
  3. retrieve and read several pages;
  4. synthesize an answer, citing some of the sources.

So classic SEO fundamentals — being crawlable, indexed and relevant — still matter. What changes is how content is selected and quoted within the answer.

Practices that help

Be the source of facts. Original data, research, clear definitions and precise statistics are more likely to be cited than generic commentary.

Answer questions directly. Put a clear, self-contained answer near the top of a section, then elaborate. AI systems extract passages; make yours easy to quote.

Structure content. Descriptive headings, short paragraphs, lists, tables and FAQs help both people and machines.

Cite your own sources. Linking to primary sources signals reliability — and matched the GEO paper’s findings.

Show expertise and trust. Author information, editorial policies, dates and transparent sourcing support credibility, consistent with Google’s guidance on helpful, people-first content.

Keep content current. Update dates and facts; AI systems prefer fresh information for time-sensitive topics.

Make it technically accessible. Fast pages, server-rendered HTML, structured data, sitemaps and a robots.txt that does not block the AI crawlers you want to reach you. Some sites publish an llms.txt file — a proposed convention summarizing a site for language models.

Build reputation beyond your site. AI systems learn about brands from reviews, news coverage, forums and reference sites.

What to avoid

  • mass-producing thin AI-written pages;
  • hidden text or prompt-injection tricks aimed at AI crawlers, which can be treated as spam or manipulation;
  • exaggerated claims that reduce trust.

Measuring results

Track referral traffic from AI assistants and AI search features in analytics, monitor brand mentions in AI answers for important queries, and watch search console data — while accepting that some value (brand awareness from citations without clicks) is hard to measure.

A quick audit checklist

  • Can the main answer to each target question be found in the first paragraph of a section?
  • Does each important page cite primary sources and include specific figures or dates?
  • Are authors, dates and editorial standards visible?
  • Is the page server-rendered and fast, with structured data where relevant?
  • Does robots.txt allow the search and AI crawlers you want to reach you?
  • Is the content updated when facts change?

What to watch next

  • Analytics tools that report visibility in AI answers.
  • Changes in how AI search features cite and link sources.
  • Publisher licensing and compensation for AI answers.
  • Standards for signaling content usage preferences to AI crawlers.

Key takeaways

  • Generative engine optimization aims to have content cited in AI-generated answers.
  • Research found citations, quotations and statistics improved visibility in AI answers in experiments.
  • Clear, direct, well-structured and current content is easiest for AI systems to quote.
  • Technical accessibility and reputation beyond your site still matter.
  • Mass-produced thin content and manipulation tricks backfire.

Frequently asked questions

Is GEO different from SEO?

It builds on SEO fundamentals — crawlability, relevance, quality — but focuses on being selected and quoted within AI-generated answers rather than only ranking in a list.

Should I block AI crawlers?

It depends on your goals. Blocking prevents use of your content but also reduces the chance of being cited. Many publishers make different choices for training versus search crawlers.

What is llms.txt?

A proposed convention: a Markdown file at a site’s root that summarizes key content and links for language models. Support varies across AI systems.