GEO glossary
Plain-language definitions for the terms that come up in GEO and AI search.
GEO (Generative Engine Optimization)
The practice of making sure a brand is one of the names an AI engine like ChatGPT, Claude, or Gemini actually recommends in a generated answer — the equivalent of ranking #1, but for AI-written answers instead of a search results page.
LLM (Large Language Model)
The type of AI model behind ChatGPT, Claude, and Gemini — trained on large amounts of text to generate answers to a prompt, rather than retrieving and ranking existing pages the way a search engine does.
RAG (Retrieval-Augmented Generation)
A technique where a model runs a live search or lookup before answering, then generates its response using what it found — as opposed to answering purely from what it memorized during training.
Grounding
When an AI answer is tied to specific, verifiable sources it retrieved (rather than generated from memory alone) — a grounded answer can cite exactly where a claim came from.
Cited source
A domain or page an AI engine actually referenced while forming an answer. The closest thing GEO has to a concrete to-do list — it names the exact sites worth having a presence on.
Visibility score
A single 0–100 number blending how often a brand is mentioned, how early it appears in the answer, and the sentiment of the mention — tracked over time so it reads as a trend, not a one-off snapshot.
Mention rate
The share of tracked prompts where a brand shows up at all in the AI's answer, regardless of position or tone.
Hallucination
A confident-sounding claim an AI model generates that isn't actually true or verifiable — in a GEO context, often a single engine naming a brand with specific detail that no other engine or independent source corroborates.
Corroboration
Multiple AI engines independently landing on the same answer about a brand. Strong corroboration is a much more reliable visibility signal than one engine confidently saying something the others don't recognize.
AI Overview
Google's AI-generated summary shown above traditional search results for many queries — synthesizes an answer from multiple sources instead of just linking to them.
Answer engine
Any AI product whose core interaction is a generated answer rather than a list of links — ChatGPT, Claude, Gemini, and Google's AI Overviews are all answer engines in this sense.
Knowledge cutoff
The date after which a model's training data stops — anything a brand does after that date won't be part of the model's memorized knowledge unless the model also retrieves live information at answer time.
Structured data
Content organized so a fact is stated plainly and unambiguously (a price, a spec, a claim) rather than buried in narrative marketing copy — easier for both search crawlers and AI retrieval to fetch and quote correctly.
Sentiment
The tone of an AI answer toward a brand when it's mentioned — positive, neutral, or negative — tracked alongside mention rate and position since being mentioned often isn't the same as being recommended.
Position
Where a brand lands among the other names an AI answer recommends — first mentioned generally reads as the strongest recommendation, similar in spirit to rank #1 in traditional search.
Want to see these show up in a real report?
Explore live demo →