The practice of improving a piece of content’s ability to answer users’ questions clearly and appear in answer engines. It relies in particular on useful, accurate and well-structured content that matches search intent. The term is still recent and its definition varies between practitioners.
A set of practices designed to improve the visibility of a brand, content or source in answers generated by AI systems such as ChatGPT, Gemini or Perplexity. The boundaries between GEO, AEO and SEO remain unclear and are likely to evolve. Google considers optimisation for its generative features to be part of SEO. (1)
The ability of a business, brand or piece of content to be understood, mentioned, cited or recommended in answers generated by AI systems and answer engines.
A system that aims to provide a concise answer to a question directly, rather than simply presenting a list of links to explore.
AI-generated answers displayed directly in Google Search, accompanied by links to sources where users can learn more. (1)
Google’s conversational search experience, designed in particular for complex questions, comparisons and research involving multiple steps. It can run several related searches to build an answer.
The presence of a website or page as an identifiable source in an AI-generated answer, usually with a link to the source.
The presence of a company, brand, product or person’s name in an AI-generated answer, with or without a link to a source.
A search in which the user obtains enough information directly from the search or assistant interface without having to visit another website.
A search method in which the user asks complete questions, adds context and progressively refines their request through dialogue, rather than entering only a few keywords.
A technique that allows an AI system to search external sources before generating its answer. This makes it possible to use information that is more recent or specific than the model’s training data. Google says it uses RAG approaches for some generative search features.
The process by which an AI links its answers to verifiable external sources to ensure their factual accuracy.
A technique used by some generative engines that involves breaking a question down into several related searches to gather the information needed for an answer. Google describes this approach for AI Overviews and AI Mode.
A clearly identifiable object — a company, person, product, place, organisation or concept — to which different pieces of information can be linked. Helping search engines understand a website’s entities correctly reduces ambiguity.
Information added to a page’s code to describe entities, properties and relationships explicitly: a company, product, article, event, review and so on. Schema.org provides the most widely used vocabulary for this purpose.
A commonly used format for embedding structured data in a web page. It makes it possible to express Schema.org information in a machine-readable form.
A proposal for a file placed at the root of a website to provide AI systems with a structured selection of useful resources and information. It remains a proposal rather than a universally adopted standard. Google Search currently says it ignores llms.txt, although other tools may use it.
An AI-based system capable not only of producing an answer, but also of planning and carrying out actions to achieve a goal: finding information, comparing products, calling an API, making a booking or triggering a process.
The ability of a website, product or system to be correctly understood and used by AI agents through accessible information, usable data and, when necessary, interfaces that allow actions to be performed.
An automated program used by an AI provider to browse web pages. Different crawlers may serve different purposes, including search, training and information retrieval.
A file used to tell robots which content on a website they are or are not allowed to crawl. It can be used to manage access for certain AI crawlers.
A protocol that enables AI applications and agents to access external tools, resources and systems in a standardised way.