An AI Visibility Glossary: 25 Terms, Plainly Defined

The terms

AEO (Answer Engine Optimization). Making your content the thing an engine quotes when it answers a question. The emphasis is on being the answer, not on ranking.

AI crawler. A bot an AI company runs to fetch web pages, for training or live retrieval. They identify themselves by user agent and most respect robots.txt.

AI Mode / AI Overview. Google's generated-answer surfaces. Overviews sit above regular results; AI Mode is a full conversational interface.

Answer engine. Any system that returns a composed answer rather than a list of links: ChatGPT, Perplexity, Claude, Gemini, Copilot, and voice assistants built on them.

Answer-ready page. A page with a direct, quotable statement near the top that answers one specific question, followed by supporting detail.

Citation. A link or source reference an engine attaches to part of its answer. Being cited is the observable unit of AI visibility.

Context window. The amount of text a model can consider at once. Retrieved pages compete for this space, so concise pages do well.

Crawl access. Whether a given crawler can reach and read your pages. Checked per user agent against robots.txt, firewalls, and bot protection.

Entity. A thing the engine can identify as distinct: a company, product, person, or place. Engines reason about entities more than about pages.

Entity consistency. Having the same facts about an entity everywhere it appears: your site, schema, business profiles, review sites, and knowledge bases.

Fabrication. A model stating something false with confidence. For brands it is usually a retrieval gap: the model could not find the fact and guessed.

GEO (Generative Engine Optimization). The broader term for influencing what generative engines say. Often used interchangeably with AEO.

Grounding. Tying a generated answer to retrieved sources so it can be checked. Grounded answers carry citations; ungrounded ones come from memory.

JavaScript dependence. Content that only exists after a browser runs scripts. Many crawlers do not run them, so that content is invisible.

Knowledge cutoff. The date after which a model's training data stops. Anything newer must come from retrieval.

llms.txt. A proposed plain-text file at the site root summarizing what the site is, for models. Cheap to write, unevenly adopted.

Mention. Your brand named in an answer, with or without a link. Weaker than a citation but still visibility.

Position. Where you appear in an answer that names several options. First and second get most of the attention; fifth is close to absent.

Prompt set. The fixed list of questions used to measure visibility over time. Should reflect what real buyers ask.

RAG (Retrieval-Augmented Generation). The architecture where a model retrieves documents first, then writes an answer using them. Most live answer engines work this way.

Rendering. Executing a page's JavaScript to produce the final HTML. A crawler that renders sees more; most AI crawlers render little or nothing.

Schema markup. Structured data in JSON-LD describing entities on a page. Organization, Product, FAQ, and Article types are the ones engines parse most reliably.

Sentiment. How an answer characterizes you: recommended, mentioned neutrally, or named as a caution. Tracked alongside presence.

Share of answer. The percentage of answers in your prompt set, across engines, in which you appear. The core metric for the work, tracked weekly.

Third-party source. Any page about you that you do not control: Reddit, G2, YouTube, editorial lists, forums. Engines tend to weigh these more heavily than your own site.

How the terms fit together

An answer engine receives a question from your prompt set. It checks crawl access, retrieves pages it can read without rendering, prefers answer-ready pages and trusted third-party sources, resolves entities using schema and entity consistency, and composes a grounded answer with citations. Share of answer, position, and sentiment are what you measure at the end. A fabrication usually means a gap earlier in that chain.

At AEO GEO Labs every engagement uses these terms the same way, from the audit through the weekly report, so a client in month six knows exactly what each number means.

Use the list

If a term here is new to you, that is probably the layer you have not checked yet. The free tools map onto several: the ai-crawler-checker for crawl access, the javascript-visibility-checker for rendering, the answer-ready checker for quotability, and the schema and llms.txt generators for the entity layer. For the full chain measured end to end, AEO GEO Labs starts with an audit that reports on each of these in plain language.