AIEO AB

Guide · updated August 2026

Generative Engine Optimization (GEO): how to get cited by AI answers

Generative engine optimization is the practice of preparing content so that generative search systems can retrieve it, understand it and quote it inside an answer. This guide explains what changes compared to classic SEO, and gives you a checklist you can run against your own page for free.

What is generative engine optimization?

Generative engine optimization (GEO) is the process of structuring, sourcing and publishing content so that AI answer engines — ChatGPT Search, Perplexity, Google AI Overviews, Gemini, Claude and Copilot — retrieve it, trust it and cite it in generated answers. Where SEO optimises for a ranked position in a list of links, GEO optimises for inclusion in the answer itself.

GEO vs SEO: what actually changes

Search engines return a ranked list; generative engines assemble a single answer from a handful of retrieved passages. That changes which signals matter. Crawlability and speed still count, but the deciding factors become whether a model can parse your page without executing JavaScript, whether the claim it needs sits in a short, self-contained passage, and whether your entity is unambiguous enough to be named.

The unit of optimisation shrinks from the page to the passage. A 2,000-word page with the answer buried in paragraph fourteen loses to a page with a clear heading, a direct one-sentence answer, and a table of supporting facts.

  • SEO target: rank in the top ten links. GEO target: be one of the cited sources.
  • SEO unit: the page. GEO unit: the quotable passage.
  • SEO gatekeeper: the crawler and the index. GEO gatekeeper: the retriever and the model's trust in your source.
  • SEO metric: position and clicks. GEO metric: citation share and brand mentions inside answers.

How generative engines pick sources

Most answer engines run retrieval before generation: they issue searches, fetch a small set of pages, and feed extracted text to the model. Anything the fetcher cannot read — client-rendered content, text inside images, gated pages, PDFs without text layers — never reaches the model, no matter how good it is.

Once text is retrieved, the model favours passages that are specific, attributable and consistent with other sources. Concrete numbers, named entities, dates and explicit statements of fact survive summarisation; vague marketing prose does not.

Make your content machine-readable

Serve meaningful content in the initial HTML response. Give each page one descriptive H1 and a logical heading hierarchy, so a retriever can slice the page into sections. Add Schema.org JSON-LD — Organization, Article, FAQPage, Product — so entities and relations are stated explicitly rather than inferred.

Keep the facts a model would need in plain text near the top: what the thing is, who publishes it, when it was updated. Tables of specifications and definitions are unusually easy to quote correctly.

Let AI crawlers in

Check robots.txt for the agents that matter: GPTBot and OAI-SearchBot (OpenAI), ClaudeBot (Anthropic), PerplexityBot, Google-Extended, Applebot-Extended, CCBot and Bytespider. Many sites block these by accident through a default rule or a CDN bot-protection setting, then wonder why they are never cited.

A /llms.txt file at the domain root gives models a short, plain-markdown brief on what your site is and where the important pages live. It is not an official standard, but it costs nothing and is trivially parseable.

Earn trust the model can verify

Generative engines lean on corroboration. Name the author, publish and maintain a visible last-updated date, cite primary sources, and make sure third-party profiles — registries, directories, Wikipedia-class references — agree with what your own site says. Contradictions push a model toward the source it considers safer, which is rarely yours.

Measure GEO results

Classic rank tracking will not tell you whether you were cited. Track prompts instead: ask the assistants your customers use the questions your customers ask, and record whether you appear, how you are described, and which URL is linked. Watch referral traffic from chatgpt.com, perplexity.ai and gemini.google.com in your analytics as a second signal.

GEO checklist

  1. 01Meaningful content present in server-rendered HTML (works with JavaScript disabled).
  2. 02One descriptive H1, plus a clean H2/H3 hierarchy that maps to real questions.
  3. 03A direct, self-contained answer within the first two sentences of each section.
  4. 04Schema.org JSON-LD for Organization and the page type; FAQPage where there are real Q&As.
  5. 05Title and meta description that state the topic literally, not cleverly.
  6. 06robots.txt explicitly allowing GPTBot, OAI-SearchBot, ClaudeBot, PerplexityBot, Google-Extended and CCBot.
  7. 07/llms.txt with a short brief and links to key pages.
  8. 08Visible author, publish date and last-updated date; matching dateModified in JSON-LD.
  9. 09Facts stated as facts: numbers, dates, names, units — not adjectives.
  10. 10Canonical URL, correct lang attribute, descriptive alt text on informative images.
  11. 11Consistency between your site and third-party profiles about your entity.
  12. 12A sitemap.xml that lists every page you want retrieved.

Check your page against these rules

The scan below fetches a page the way an AI crawler does — no JavaScript, no login — and grades fifteen of the signals described above, with a fix suggestion for each one.

AI visibility quick scan

Paste any public page. We fetch it the way an AI crawler does — no JavaScript, no login — and grade how readable, structured and citable it is. Nothing is stored.

Provided free of charge, as-is and without any warranty. The scan is an automated, rule-based indication only — not an audit, and not legal, technical or commercial advice. AIEO AB accepts no liability whatsoever for the results or for any action taken on the basis of them. Only scan pages you own or are entitled to test.

GEO questions

Is GEO the same as AEO or AIEO?

Broadly yes. Generative engine optimization (GEO), answer engine optimization (AEO) and AI engine optimization (AIEO) all describe optimising for AI-generated answers rather than ranked links. GEO is the term most used in research and industry writing; the practices behind the three labels overlap almost completely.

Does GEO replace SEO?

No. Retrieval still runs on search infrastructure, so an unindexed page is invisible to both. GEO is an additional layer: technical readability, explicit structure, verifiable facts and crawler access on top of a healthy SEO baseline.

How long does GEO take to show results?

Indexed pages can start appearing in answers within days once retrieval picks them up, while the knowledge baked into a model's weights lags by months. Fixes to crawler access and structured data tend to show up fastest.

Does blocking AI crawlers protect my content?

It prevents citation as well as training use. If discovery through AI assistants matters to you, allow the search-oriented agents (OAI-SearchBot, PerplexityBot) even if you choose to block training-only ones.

What is the single highest-impact GEO fix?

Serving your real content in the initial HTML response. Everything else is worthless if the retriever receives an empty shell.