Glossary

AEO — Answer Engine Optimization

Optimizing to be the answer rather than a result. Used interchangeably with GEO by most practitioners; where people distinguish them, AEO leans toward direct-answer formats and GEO toward generative synthesis.

AI Overviews

Google's AI-generated summaries at the top of search results. Consequential because they sit above the classic results on queries that used to send you clicks. Tracked as its own engine.

AI Mode

Google's conversational search surface, distinct from AI Overviews and tracked separately.

Answer

One capture of one prompt on one engine, on one day: the text as rendered, its citations, its fan-out queries and any placements. The atomic unit everything aggregates. See Core concepts.

Citation

A source an engine displayed with an answer. Quoted citations are cited inline against a specific claim; listed citations appear in the sources panel beside it. Distinct from a mention: a citation is a source, a mention is your brand appearing in the text.

Citation index

How often a brand's own domain appears as a source, as opposed to how often the brand is named. See Citations and sources.

Confidence interval

The band around a measured rate, at 95% by default. Produced by asking each prompt several times per engine per day and measuring the spread. A rate without one is a point estimate pretending to be a fact.

Crawler (AI)

A bot fetching pages for an AI system. Training crawlers collect data to train models; search crawlers fetch pages to answer live questions. Blocking the second removes you from AI answers; blocking the first does not.

Engine

An answering surface Geonimo measures — ChatGPT, Google AI Overviews, Google AI Mode, Perplexity — captured from the real consumer interface. Distinct from a provider, which is the model family behind it.

Entity

Anything tracked inside answers: your brand, a competitor, or a product.

Fan-out query

A machine-written search an engine issues behind a user's question. One conversational prompt typically produces several. See Query Fan-out.

GEO — Generative Engine Optimization

Optimizing to be found, named and recommended inside AI-generated answers, as opposed to ranking in a list of links.

Grounding

Answering with live web search rather than from training memory. Every captured answer records whether it actually surfaced sources, reported as the Grounded Answers KPI. See AI engines.

Hallucination

A confidently stated falsehood. In this context usually a wrong price, a discontinued feature, or a company attribute the engine invented.

Intent (prompt)

Whether a prompt measures visibility — a category question that may not name you — or perception — a question about you by name. The two are never mixed in one metric. See Prompts.

llms.txt

A file convention describing your site to AI systems — what it is about, which pages matter. Young standard, cheap to adopt. Generator in Free tools.

Mention

A tracked entity detected inside an answer, with its position, count and snippet.

Noise floor

The threshold a change must clear before it is marked significant. Wider than the confidence interval on either endpoint, because comparing two estimates compounds their uncertainty.

Placement

A shopping or ad slot rendered alongside an AI answer. Measured as share of slots rather than share of answers. See Ads & Shopping.

Position

Order of first appearance among tracked entities in an answer. First named is position 1. Lower is better.

Prompt

A question tracked on your behalf, asked of every engine every day. The unit of measurement and of cost.

RAG — Retrieval-Augmented Generation

The general technique behind grounded answers: retrieve documents, then generate from them. AI search engines are large-scale RAG systems, which is why what you publish matters as much as what the model learned.

Recommendation strength

Whether a mention lists you, recommends you, or warns against you. The commercially important distinction a mention count cannot see. Arrives with the enrichment pass.

Reasoning mode

Which mode of a surface produced an answer — default, instant or thinking. Modes retrieve and cite differently, so the dimension is carried rather than flattened.

Sentiment

Net tone of your mentions, -100 to +100. Measures how you are described, not the answer's overall mood.

Share of slots

Your share of the entries inside shopping and ad units. Used instead of an answer-level rate because a carousel of ten is ten chances to be seen.

Share of Voice (SoV)

Of the answers mentioning any tracked entity, the share mentioning yours. The competitive metric. See Share of Voice.

Source

A specific URL cited in answers, with a title and first/last seen dates. Sources roll up to domains.

Structured data

Machine-readable markup (JSON-LD schema) describing page content. FAQ markup on question-shaped pages is consistently the highest-return version for citability.

Visibility

Share of tracked answers that mention you, measured on unbranded prompts only. Am I in the room? See Visibility score.

Workspace

A team account: the brand, prompts, members and subscription. Everything in the app is scoped to one.