AI visibility glossary

Answer Engine Optimization (AEO)

Answer Engine Optimization (AEO) is the practice of getting a brand named inside the answer an AI engine writes, rather than ranked in a list of links below it. It matters because a buyer who asks ChatGPT or Google’s AI Overview for a recommendation usually acts on the two or three brands the answer names, and never sees the rest.

AEO and SEO are not alternatives. Search ranking still feeds the engines their source material; AEO is about whether the engine names you when it summarises that material.

Answer engine

An answer engine is a system that responds to a question with a written answer instead of a list of links. ChatGPT, Claude, Perplexity and Google’s AI Overview are the four that matter for most buying questions today.

The distinction from a search engine is what it returns, not what it is built on. Google is both: the ten blue links are search, the AI Overview above them is an answer engine.

AI visibility

AI visibility is how often, and how prominently, a brand is named in the answers AI engines give to buying questions in its category. It is measured by asking the engines real buyer questions and recording which brands come back, not by checking whether a page is indexed.

AI Overview

An AI Overview is the AI-written summary Google places above its normal search results for some queries. It cites the pages it drew from, so it is one of the few answer surfaces where a brand can see exactly which sources produced the answer.

It does not appear for every query, and whether it appears at all can change week to week for the same question. A brand can rank first in the blue links and be absent from the Overview directly above them.

Citation

A citation is a source an AI engine links to in support of an answer. Citations decide which version of the facts an answer carries: an answer built from a brand’s own pages repeats that brand’s pricing and positioning, while one built from third-party round-ups repeats whatever those round-ups say.

An owned citation points at the brand’s own domain. A brand can be named frequently and still own almost none of the citations about it, which usually means AI is describing it from someone else’s summary.

Share of voice

Share of voice, in an AI context, is a brand’s share of all brand mentions across a set of answers. If ten answers name brands thirty times in total and three of those name you, your share of voice is 10 percent.

It is a different measure from how often you appear. A brand named once in every answer has high presence and low share of voice if each of those answers names six competitors alongside it.

Consideration set

A consideration set is the short list of brands an answer actually puts forward when a buyer asks for a recommendation. Being outside it is different from being unknown: the engine may describe a brand accurately when asked directly and still leave it out when asked what to buy.

Grounding

Grounding is the step where an AI engine fetches live sources and writes its answer from them, rather than from what it learned during training. A grounded answer can change the day a source page changes; an ungrounded one reflects whatever the model absorbed months earlier.

Retrieval-augmented generation (RAG)

Retrieval-augmented generation (RAG) is the technique behind grounding: the system searches for relevant documents first, then asks the language model to answer using them. It is why a brand’s own pages can influence an answer without the model ever having been trained on them.

Brand hallucination

A brand hallucination is an AI answer that states something about a brand which is not true: a price that was never charged, a feature that does not exist, or a company detail that belongs to someone else. It happens most often when the engine has little first-party material to work from and fills the gap from similar-sounding sources.

Buyer prompt

A buyer prompt is a question a real prospective customer would type, phrased the way they would type it. "Best AI visibility tool for Indian SaaS" is a buyer prompt; "AnswerTrace" is not, because a brand appearing in the answer to its own name proves nothing about whether buyers can find it.

This distinction decides what a visibility measurement is worth. Counting brand-name lookups alongside category questions inflates the result, sometimes dramatically.

Entity

An entity is a thing an AI engine can identify and hold facts about, such as a company, a product or a person, as distinct from a string of characters that happens to appear on a page. A brand that reads as an entity gets a consistent description across answers; one that does not gets confused with similarly named things.

Wikipedia, Wikidata and a brand’s own structured data are the sources most often used to pin an entity down.

llms.txt

llms.txt is a proposed plain-text file at the root of a site that lists the pages a site owner would most like an AI engine to read. It is a convention rather than a standard, and engines are free to ignore it.

Treat it as cheap and unproven. It costs an hour to publish and no engine has committed to honouring it, so it is worth doing only after the pages it would point at are worth pointing at.

Striking distance

A striking-distance keyword is one a site already ranks for on Google, but just below the point where clicks arrive, conventionally positions four to fifteen. They matter because moving a page from eleventh to fifth is usually far less work than ranking a new page at all.