30+ terms defined, plain English. The category itself is new - most teams need shared vocabulary before they can ship a program.
Optimizing content so answer engines (Perplexity, Google AI Overviews, ChatGPT Search) quote it as the direct answer to a query.
The umbrella term for optimizing for any AI-powered search surface - including ChatGPT Search, Perplexity, Google AI Overviews, and AI Mode.
The practice of optimizing brand and content so generative AI engines (ChatGPT, Perplexity, Gemini, Claude, Grok) mention, recommend, and cite you.
Optimizing for Large Language Models - making sure ChatGPT, Claude, Gemini, Perplexity, and Grok know about, cite, and recommend your brand.
Google's full conversational search experience - an alternative to classic SERPs where AI answers the query end-to-end.
Google's generative-AI answer box at the top of search results, powered by Gemini. Now appears for more than half of qualifying US queries.
OpenAI's live-retrieval mode inside ChatGPT - pulls real-time web results, cites them, and answers in the chat surface.
Anthropic's LLM family (Opus, Sonnet, Haiku). Available via Claude.ai, the Anthropic API, and embedded in Notion, Slack, Quora, Cursor, and more.
Google's multimodal LLM family. Powers Gemini app, AI Overviews, AI Mode, Workspace AI, and the Vertex AI / AI Studio APIs.
xAI's LLM, with native access to real-time X (formerly Twitter) data. Powers Grok-on-X and Grok.com.
A dedicated answer engine that retrieves the web in real time and synthesises answers with inline citations.
How consistently many independent sources describe a brand the same way. The single biggest factor in whether an LLM names a brand by default.
How easily an LLM can lift a clean, summarisable answer from your page. High extractability dramatically increases citation rate on grounded surfaces.
Augmenting an LLM's answer with retrieved evidence so the model can cite verifiable sources, rather than relying on training memory alone.
A technique where an LLM retrieves external documents at inference time and uses them to ground its answer. Powers Perplexity, ChatGPT Search, and AI Overviews.
JSON-LD or microdata that labels your content for machines. FAQ, Article, Product, HowTo, and Organization schema all improve LLM extractability.
The dataset an LLM was trained on. Brands that appear frequently and consistently in the training corpus are recalled by name in answers, with no live retrieval required.
Percentage of cited sources in an LLM answer that come from your domain. Only meaningful on grounded surfaces (Perplexity, ChatGPT Search, AI Overviews).
The percentage of prompts in a defined panel where an LLM names your brand. The headline metric of LLM SEO programs.
Where your brand appears in an LLM's ordered recommendation set - first paragraph (best), middle, or final mention (weakest).
The tone with which an LLM describes your brand - positive, neutral, negative - when it does mention you.
Your mention rate as a percentage of the total mentions of you plus your named competitors across LLM answers.
Searches that end inside the search surface without a click to an external site. Now 58% of US searches.
Anthropic's web crawler for Claude training and (in some configurations) retrieval.
An opt-in agent that lets Google use your content for Gemini training. Separate from Googlebot, which is required for classic search.
OpenAI's crawler for ChatGPT training data. Identified by the user agent "GPTBot".
A proposed standard file at the root of a site that tells LLM crawlers what content is available, in what format, and how to use it.
OpenAI's live-retrieval crawler for ChatGPT Search. Separate from GPTBot and must be allowed independently.
Perplexity's live-retrieval crawler. Blocking it removes you from Perplexity citations.
The decades-old file telling web crawlers what they can fetch. Used to allow or block GPTBot, ClaudeBot, PerplexityBot, Google-Extended, and others.
AI search is a new category and the language is still settling. Two teams will use "GEO", "AEO", and "LLM SEO" for slightly different things; vendors will pick whichever sounds best in a deck. This glossary picks one definition for each term and sticks with it, so internal conversations and external content stop talking past each other.
Suggestions or corrections: hello@livesov.com.
Track ChatGPT, Perplexity, Claude, Gemini and Grok. 7 day free trial, no credit card.