How to get cited by ChatGPT and AI Overviews: what the engines actually say

Are GEO and AEO different from SEO?

Not according to Google. Its guide to optimising for generative AI features, last updated in July 2026, says that optimising for generative AI search is optimising for the search experience, “and thus still SEO”. Other engines have their own technical rules, but the principle holds: substance beats shortcuts.

Generative Engine Optimization (GEO)

Work that raises the chance that a generative engine, such as ChatGPT, Perplexity, Copilot or Google's AI Overviews, uses and cites your pages in its answers. The term comes from a paper by researchers at Princeton and IIT Delhi, accepted at KDD 2024. Answer Engine Optimization (AEO) is the same goal seen from the direct-answer side.

The same Google guide clears away several myths. You do not need to chop text into small chunks for AI, there is no special schema.org markup for generative features, and files such as llms.txt are not used by Google Search. What it does ask for is an original point of view grounded in first-hand experience, which it calls non-commodity content.

How do ChatGPT, Perplexity and Copilot choose their sources?

They pick from pages their search systems can reach and judge relevant to the question. So the first check is technical: if an engine's search crawler is blocked, that engine cannot cite you, however good the page is.

ChatGPT search

OpenAI runs three separate agents, as its crawler documentation explains: OAI-SearchBot for search, GPTBot for training foundation models, and ChatGPT-User for visits a user triggers. Sites that opt out of OAI-SearchBot are not shown in ChatGPT search answers, except as plain navigational links. Each setting is independent, so you can stay in search while opting out of training. OpenAI says robots.txt changes take about 24 hours to reach its systems.

Perplexity

PerplexityBot surfaces and links sites in Perplexity's results and, per the company, is not used to train foundation models. Perplexity-User fetches a page when someone asks a question and links it in the answer; because a person requested it, it generally ignores robots.txt. Changes can take up to 24 hours, according to Perplexity's crawler docs.

Copilot and Bing

Microsoft now reports Copilot citations inside Bing Webmaster Tools. When it launched the AI Performance dashboard, Bing listed what helps a page get used as a source: clear headings, tables and FAQ sections, claims backed by examples, data and cited sources, fresh content, and text, images and video that describe the same things consistently.

What does a page need to appear in Google's AI Overviews?

It must be indexed and eligible to be shown with a snippet. In its documentation on AI features and your website, Google says there are no additional technical requirements, and that meeting these does not guarantee crawling, indexing or serving.

One mechanism explains why AI Overviews sometimes cite pages that do not rank first for the original query: query fan-out. The system runs several related searches across subtopics and sources to assemble its answer, so a page that nails one subtopic can earn a link without winning the head query.

Snippet controls apply too. nosnippet, data-nosnippet and max-snippet limit how much of your text Google can show, including in AI answers. They help you keep parts of a page out; they do not help you get in.

How do you write a page that AI engines cite?

Answer straight away and prove what you claim: a direct answer up top, then detail, sourced figures, quotations and your own experience. In the Princeton and IIT Delhi GEO study, the strongest methods were exactly those: citing sources, adding quotations and adding statistics.

Two findings matter for a young site. Keyword stuffing did not help and actually scored below the unoptimised baseline. And lower-ranked pages gained the most: citing sources lifted visibility by 115% for pages in fifth position and cut it by 30% for pages in first. It is an experiment on test engines, not a guarantee of how ChatGPT behaves today, but it points the same way as Google's guidance.

How can you tell whether AI engines cite you?

Only partly, for now. Google and Bing publish official numbers; ChatGPT and Perplexity do not. Google launched its Search Console generative AI performance reports on 3 June 2026 and rolled them out to all sites by 31 August 2026. They show impressions in features such as AI Overviews and AI Mode, broken down by page, country, device and date; clicks are not among the listed metrics.

Bing's AI Performance dashboard, in public preview since February 2026, counts how often your pages are cited in Copilot and Bing's AI summaries, which URLs, and the grounding queries the AI used to retrieve them. To keep what the engines read up to date, Bing recommends IndexNow.

For ChatGPT and Perplexity, the method is manual: a fixed list of your customers' real questions, rerun monthly, noting who gets cited and with which page. Keep citations and traffic as separate metrics, because one does not imply the other.

Which shortcuts backfire?

• Hundreds of pages for question variants: Google treats this as scaled content abuse, even when the stated aim is AI answers.
• Blocking the wrong crawler: teams that want to stop model training can end up blocking search by mistake. GPTBot and OAI-SearchBot are different agents.
• Relying on FAQ markup for extra space in results: according to Google's documentation updates log, that rich result has not been shown since 7 May 2026. Visible FAQs still help readers and engines.
• Paying for guaranteed citations: Google points out that third-party tools have no access to its internal data and cannot guarantee anything.
• Changing dates without changing content: Google lists it as a sign of pages made for search engines rather than for people.

If you are starting from scratch, read our SEO guide for startups first, because GEO and AEO rest on the same foundations. And if you are wondering whether it is time to invest at all, see when a startup should start SEO.