GEO glossary: the core terms of AI visibility

This glossary defines, one by one and briefly, the terms used when people talk about AI visibility and GEO (generative engine optimization). GeoMono is a platform that measures how visible brands are in AI answers; the measurement terms here mean exactly what they mean in our measurement and our reports. Every term has its own heading and link, so you can share a term directly.

How to read this glossary

The first paragraph under each term is the definition itself: one or two sentences that can be read without further context. The second paragraph, where there is one, explains why the term matters in practice or how it differs from a term it is often confused with. The terms fall into three groups: core concepts, measurement concepts and technical concepts. The whole measurement is written up on our method page.

The terms most often confused

Some terms in this glossary sit very close to each other and are often confused. The table below gives the difference in one line; each term's full definition is under its own heading.

Core concepts

GEO (generative engine optimization)

GEO is the work of making sure a brand, product or organisation is described accurately in the answers AI engines generate, and is part of the answer to a buyer's question.

The term spread with the paper "GEO: Generative Engine Optimization", published in 2023 and presented at KDD in 2024. SEO is about ranking in search results; GEO is about the content of the AI answer. A longer explanation is on the what is GEO page.

AI visibility

AI visibility is how often, and how, a brand is mentioned in the answers to questions buyers ask AI engines. It is not a single number but a frequency measured separately for each engine.

Answer engine

An answer engine is an AI system that produces a written answer to a user's question instead of a list of links. ChatGPT, Gemini, Claude and Perplexity are answer engines; Google AI Overviews and AI Mode are answer surfaces inside Google Search.

AI Overviews

AI Overviews is the AI-generated summary answer Google shows above its search results; its Turkish name is AI Bakışı. It gives supporting links to the pages the answer rests on, chosen, according to Google's AI features documentation, from pages in Google Search's index.

According to the Google Türkiye Blog, AI Overviews and AI Mode rolled out gradually in Türkiye from 18 February 2026.

AI Mode

AI Mode is the conversational AI search experience in Google Search, which proceeds with follow-up questions. It is a separate surface from AI Overviews and can give different answers and links to the same question.

Measurement concepts

Mention

A mention is a brand's name appearing in an AI answer, meaning that brand. An answer about another company or person with the same name is not a mention of the brand.

Citation

A citation is a site or page being among the links an AI answer rests on. An engine can cite a page and never say the name of the brand that owns it.

Mentions and citations are separate events and should be counted separately. We cover the difference with examples in brand mentions and source citations.

Question that does not name the brand

A question that does not name the brand is a question written the way a buyer would ask an AI engine, without the measured brand's name. "Which accounting software suits small businesses in Istanbul?" is one; "Is X software any good?" is not.

Visibility is measured with these questions. In a question that contains the brand's name, the engine repeats the name anyway; that answer reflects the question, not whether the brand is known.

Visibility rate

The visibility rate is the number of measurable questions that do not name the brand in which the brand was mentioned, in one engine. It is written as a pair, such as "mentioned in 12 of 40 questions", and given separately for each engine.

Merging the engines' rates into one number hides a strong result in one engine behind a weak result in another. That is why the rate is read engine by engine.

Unmeasured answer

An unmeasured answer is one that could not be measured because the engine returned an error, timed out or its answer could not be read. It does not count as "brand not mentioned": it is left out of the rate, reported separately and retried.

Counting an unmeasured answer as "not mentioned" reports a provider outage as a finding about the brand. A report that does not say how many answers could not be measured looks more certain than it is.

Question set

A question set is the versioned list of questions registered for a brand before measurement. The same set is asked again in every measurement, so two measurements can be compared.

Monthly measurement

The monthly measurement asks the brand's registered question set again in five engines on a regular schedule. The visibility rate and the comparison with the previous measurement are calculated only from this measurement.

Quick check

A quick check is a short measurement with 5 to 20 questions the brand writes itself for one topic. It shows who AI mentions on that topic and which sources the answers rest on; it gives no rate, score or trend.

A few questions do not tell a stable rate: the same question can get different answers on different days. So a quick check's figures are a sample and do not replace the monthly measurement.

Full report

The full report is the report that comes out of the monthly measurement: the visibility rate for each engine, the comparison with the previous measurement, the rivals mentioned in the brand's place, the sources the answers rest on, and what to do.

Namesake

A namesake is another company, product or person with the same or a very similar name to the measured brand. A name appearing in an answer does not prove the brand was meant; if namesakes are not filtered out, the rate is inflated.

Next step

Is your site open to AI crawlers?

Enter your domain. The free scan measures whether AI crawlers can reach your site and whether your pages are quotable. No signup, no card, no email.

Your domainScan for free

Technical concepts

Structured data

Structured data is code that describes a page's content in a standard format machines can read; it is usually written with the schema.org vocabulary in JSON-LD. It states plainly that a page is an article, a product or a company, and who published it.

Google's introduction to structured data says this markup must say the same as the content on the page. We cover why it matters for GEO in structured data and GEO. This glossary itself marks up each term as a defined term.

llms.txt

llms.txt is a Markdown file placed at a site's root that lists the site's most important pages, with short descriptions, for AI systems. It was proposed by Jeremy Howard on 3 September 2024.

It is a proposal, not a standard; how far the engines read the file is unclear. We discuss when it helps in does llms.txt work.

AI crawler

An AI crawler is an automated program an AI company uses to read web pages. Some train models, some read a page at the moment they answer a user's question.

A site owner decides which crawlers to allow in the robots.txt file. Blocking a crawler can keep that engine from citing your site; details are in the robots.txt and AI bots guide.

Web search (grounding)

Web search, also called grounding, is an answer engine searching the web before it writes an answer and resting the answer on the pages it found. Answers with a web search usually cite sources; answers without one come from the model's own knowledge.

Some engines search for every question; others decide per question whether a search is needed. So the same question can be answered with sources in one engine and without them in another.

IndexNow

IndexNow is an open protocol a site uses to notify search engines of new or changed pages immediately. Engines such as Bing and Yandex use it; Bing's index also feeds Copilot and ChatGPT's web search.

Google does not use IndexNow; new pages are reported to Google through Search Console. Details are on the IndexNow site.

Frequently asked questions

Are GEO and SEO the same thing?

No. SEO is about a page ranking high in search results. GEO is about a brand being described accurately and mentioned inside the AI answer. The two support each other: answer engines often find pages through a search index. So a sound SEO base is needed for GEO too, but it is not enough on its own: how the brand is described inside the answer has to be measured separately.

What is the difference between a mention and a citation?

A mention is the brand's name appearing in the answer. A citation is the brand's site being among the links the answer rests on. An engine can cite your site without saying your name, or say your name without citing your site. Counting the two separately is the first condition for reading visibility correctly; putting one in place of the other makes the rate look higher or lower than it is.

Why is visibility measured with questions that do not name the brand?

Because in a question that contains the brand's name, the engine simply repeats the name. If a buyer looking for a product or a firm does not know the brand's name, whether the engine recommends that brand on its own can only be seen with questions that do not name it.

Why does a quick check give no rate?

Because 5 to 20 questions are not enough to tell a stable rate; the same question can get different answers on different days. A quick check says who is mentioned and which sources are shown. The rate and the trend are calculated only in the monthly measurement, with the registered question set.