AEO & GEO Glossary — Every Term Defined

AEO (Answer Engine Optimization): Structuring content and technical signals so answer engines — ChatGPT, Perplexity, Google AI Overviews, and similar — can find, extract, and cite a brand directly in their answers.; AEO is the umbrella discipline this whole program is built around — from how a page is written down to whether an engine's retrieval step can even reach it. See how we measure it and how it's structured into Strategy or Full Execution.
GEO (Generative Engine Optimization): Optimizing content and site signals specifically for generative AI systems — the surfaces that write a synthesized answer, rather than returning a ranked list of links to click through.; GEO and AEO overlap heavily and get used interchangeably in practice; where people split them, GEO leans toward the generative/synthesis layer while AEO leans toward being the source that gets surfaced. The work is the same either way — see our methodology for exactly what we track.
AI Share of Voice: The share of relevant AI-generated answers that cite or mention a brand, out of every answer where it plausibly could have been named.; Think of it as the AI-search equivalent of organic share of voice — tracked across a defined query cluster, not a single prompt. It can rise or fall independently across engines, which is why we track them separately.
Citation: An AI engine crediting a brand as the source of an answer — a named link, an attributed passage, or the domain listed as a reference.; A citation is a stronger signal than a bare mention (the brand name appearing with no attribution) or a spot in a list of options with no credit given. Full definitions and how we log citations live on the methodology page.
Entity: A distinct, recognizable thing — a brand, person, product, or concept — that an AI system can identify and connect consistently across sources, rather than a loose string of text.; Being recognized as a clear entity, not just a keyword, is a prerequisite for consistent citation — one reason structured, consistent brand mentions across the web matter separately from any single page's copy.
RAG (Retrieval-Augmented Generation): A technique where an AI model pulls in current, external information before generating an answer, rather than relying only on what it learned during training.; RAG is what makes a brand's current content able to influence an AI answer at all — it's the retrieval step underneath grounded results. See how we account for it.
Retrievability: How easily a piece of content can be found, parsed, and pulled into an answer by an AI system's retrieval step — the prerequisite for being cited at all.; Mostly a technical concern: clean page structure, crawlable content, clear headings, and copy that answers a likely prompt near the top of the page — the same direct-answer discipline used across this glossary.
Zero-click search: A search or prompt where the person gets a complete answer directly from the engine and never clicks through to a website.; Zero-click behavior is the reason AI visibility is its own discipline now: if most relevant queries end without a click, ranking well in traditional search no longer guarantees a brand is ever seen. Being cited inside the answer becomes the only way to show up at all.
Answer engine: Any AI system — ChatGPT, Perplexity, Gemini, Google AI Overviews, and similar — that responds to a query with a synthesized answer rather than a list of links to sift through.; We track multiple answer engines by default across every engagement; see which ones and why.
Prompt: A real question a potential customer might type into an AI engine — the actual unit AI visibility gets tracked against.; A concrete example: "best CRM for a 20-person D2C team" is a prompt, not a keyword — it's phrased the way a person actually asks, which is why prompt sets look nothing like a traditional SEO keyword list.
Query cluster: A group of related prompts that share the same underlying intent, tracked and optimized for together instead of one at a time.; Clustering matters because AI engines rephrase the same underlying question many different ways — optimizing for one exact prompt in isolation misses the cluster it belongs to. Content briefs from our Strategy program are built around a full cluster, not a single prompt.
Grounded vs. ungrounded AI results: Grounded results are pulled from live, current sources at answer time; ungrounded results come only from the model's training data and can be outdated or simply wrong.; This matters directly for AI-visibility work: a brand can only earn a citation in a grounded answer, since ungrounded answers aren't retrieving anything new to cite. See our methodology for how we account for this when logging results.