GEO and AI Search Glossary: Key Terms Defined Simply

An AI system that answers questions with written text instead of a list of links, such as ChatGPT, Perplexity, Gemini, Claude and Google AI Overviews.

Generative engines either answer from what their model learned in training or search the web first and write the answer from what they find. The second kind is the one you can influence quickly, because it depends on pages that exist today.

The practice of improving how often, and how favourably, AI assistants mention and cite your business in their answers.

The term comes from a 2023 research paper by Aggarwal and colleagues, published at KDD 2024 (arXiv). See what GEO is.

The practice of shaping content so that search and AI systems choose it as the direct answer to a question.

AEO and GEO overlap heavily and are often used interchangeably. In practice both mean clear, answer-first, well-structured pages that machines can read and trust.

A Google Search feature that shows an AI-generated summary of an answer with links to supporting pages.

Google says a page must be indexed and eligible to be shown in Search with a snippet to appear as a supporting link, and that there are no additional technical requirements (Google).

A Google Search experience for AI-powered answers, covered by the same eligibility rules as AI Overviews.

To appear as a supporting link in AI Mode, a page must be indexed and eligible to be shown with a snippet, according to Google's documentation.

The question or instruction a user gives an AI assistant. In GEO tracking, a prompt is a buyer question such as "best dental clinic in Kraków".

Prompts should be written the way real customers ask, in their language and for their city. Answers can vary from one run to the next, so tracking uses many prompts and repeats them.

A link or source reference that an AI assistant shows with its answer.

Citations show which pages an engine relied on. Being cited from your own pages, and from the review and directory sites your market uses, are different signals; both are worth tracking.

Connecting an AI answer to external sources, such as web search results, so it can be checked and cited.

Search-based assistants ground their answers in pages they retrieve. Pages they cannot read cannot ground an answer.

A technique in which a system retrieves relevant documents first and then writes an answer based on them.

Search-style AI assistants work this way, which is why crawlable, clearly written pages matter.

A web robot operated by an AI company that reads web pages for AI search, user requests or model training.

Examples are OAI-SearchBot, GPTBot, PerplexityBot and ClaudeBot. They have different purposes, so you can allow some and block others. See AI crawlers explained.

The name a crawler or browser sends with each request to identify itself, such as GPTBot or Googlebot.

robots.txt rules are written against user agents. Some crawlers, such as Google-Extended and Applebot-Extended, are control tokens rather than separate crawlers.

A text file at the root of a website that tells crawlers which URLs they may fetch.

The Robots Exclusion Protocol is standardized as RFC 9309. It states a policy, not a lock: firewalls and bot protection can block crawlers regardless of it. Check yours with the AI crawler checker.

A proposed markdown file at /llms.txt that gives language models a short summary of a site and links to its key pages.

Proposed by Jeremy Howard in September 2024 (llmstxt.org). It makes no claims about rankings or traffic. See the llms.txt guide and the generator.

Code on a page that states facts about it in a machine-readable vocabulary, usually schema.org.

For a business, structured data states its name, type, address, phone and opening hours. See schema.org for the vocabulary.

A format for embedding structured data in a page as a JSON script.

JSON-LD is the most common way to add schema.org markup because it sits in one block and does not change the visible page.

A distinct thing, such as a business, person or place, that machines can identify and tell apart from others.

Search and AI systems build an understanding of your business as an entity from your site, your structured data and mentions elsewhere. Consistency helps.

Name, address and phone number: the core details that identify a local business.

Keep NAP identical on your website, maps and directories. Inconsistent details make it harder for any system to be sure which business is yours.

Your mentions divided by your mentions plus those of the tracked competitors, in the same set of AI answers.

If you are named in 20 answers and your three tracked competitors in 60 answers together, your share of voice is 20 out of 80, or 25%.

A number from 0 to 100 that summarises how visible a business is in AI answers across a set of prompts.

In Recomindo it is 50% how often you are mentioned, 25% your position when mentioned, 15% citations of your own website and 10% sentiment. It is aggregated over many prompts and engines, because single answers vary. See the AI Visibility Index.