Glossary
Every term this site uses when it measures whether AI engines cite a brand, from GEO itself down to how a single run becomes a confidence interval. Each definition traces back to the same methodology every score on this site follows.
Once a brand is selected as a source, how much it actually shapes the substance of the answer, versus sitting unused in a source list. Driven by extractable evidence density: definitions, statistics, comparisons, and structured content a model can lift directly.
A close synonym of GEO with roots in older featured-snippet and voice-search optimization practice. In practice, the same target: being cited by a system that answers directly instead of listing links.
Google's AI-generated summary that appears above the traditional search results for many queries, synthesizing an answer from several already-ranking pages and citing a handful of them directly.
A test question phrased the way a real buyer would ask it when comparing or choosing a provider, not a broad category keyword. The basis for a defensible citation measurement.
A brand or domain named, linked, or otherwise identified as a source inside an AI engine's answer to a prompt. The base unit every score on this site is built from.
The range a repeated score could plausibly fall within, computed by resampling the runs behind it. Ships with every score built from more than one run; a single run gets a snapshot label instead.
The split between chat engines (ChatGPT, Perplexity, Gemini, Claude) and Search-AI surfaces (Google AI Overviews, AI Mode, Bing Copilot). Never blended into one score.
The practice of structuring content and technical infrastructure so generative AI engines, chat assistants and AI-generated search results alike, can find, understand, and cite a brand in their answers.
Whether an AI answer draws on live web retrieval (a grounded, or retrieval-augmented, answer) versus fixed training-time knowledge. Changes what optimization work can influence, and how fast.
The declared, fixed list of questions run against each AI engine to produce a citation score. Kept stable across measurement runs so a later delta reflects the market, not a different question.
One execution of a prompt set against an engine. AI outputs are not fully deterministic, so a single run is a snapshot; several runs, or replications, form a range with a confidence interval.
Whether a brand enters the pool of sources a model draws from at all, before absorption is even possible. Driven mostly by classic authority signals: backlinks, brand recognition, third-party mentions.
The percentage of answers, across a declared prompt set, engine set, and run count, in which a brand was cited. The headline number behind every leaderboard on this site.
The honesty label for a score built from exactly one pass, not an average of repeated runs. Real data, but a photograph, not a trend line.
Comparing a score against itself over time, inside one engine. The only comparison this site treats as a real signal; comparing raw scores across two different engines is treated as noise.
Free scan, no credit card, results in minutes.