The AI visibility glossary.

This glossary defines 32 terms from the world of AI search: the everyday vocabulary, the acronyms, the crawlers, the technical plumbing and the ways visibility gets measured. Every definition opens with the definition itself, not a preamble, and each entry has its own anchor link, so you can point someone straight at a term.

• The basics: the everyday vocabulary of AI search.
• The acronyms: SEO, AEO, GEO, RAG, E-E-A-T, NAP.
• The crawlers: the bots that read your site, and what blocking each one costs.
• The plumbing: the files and markup that tell machines what your business is.
• The measurements: how being in the answer gets counted.

The basics

The everyday vocabulary. If AI search is new to you, start here.

AI visibility

AI visibility is whether AI tools such as ChatGPT, Gemini, Perplexity and Google's AI Overviews mention or recommend your business when people ask relevant questions. It differs from a search ranking: an AI assistant gives one short written answer, so a business is either named in it or absent from it, with no tenth position to occupy. We cover the whole idea in what is AI visibility.

LLM (large language model)

An LLM, or large language model, is the software behind tools like ChatGPT, Claude and Gemini. It is trained on very large amounts of text and generates its replies word by word, based on patterns in that text. An LLM does not look answers up in a database, which is why its statements about a business need checking against real sources.

Prompt

A prompt is the question or instruction a person types into an AI tool. In AI visibility work, prompts matter because different phrasings can produce different recommendations. "Best physio in Geelong" and "physio who treats runners in Geelong" may surface different businesses, so visibility is measured across many prompts rather than a single favourite one.

Knowledge cutoff

A knowledge cutoff is the date after which an AI model's training data ends. Anything published later is unknown to the model unless it retrieves live web results at the time it answers. This is why a recent change to your business can be missing from some AI answers and present in others, depending on whether retrieval was used.

AI Overviews

AI Overviews are the AI-written summaries Google shows above its ordinary results for many searches. They are generated by Google's Gemini models, draw on pages from Google's index and link to a small set of sources. Appearing in that source list is a distinct goal from ranking below it. Our guide to showing up in Google AI answers covers the practical side.

AI Mode

AI Mode is Google's conversational search experience, offered as a separate tab where the whole results page is an AI-generated answer with follow-up questions, rather than a list with a summary on top. It runs on Gemini and is available in Australia. Because it replaces the list of links entirely, being cited inside the answer is the only way to be present.

Hallucination

A hallucination is a confident but false statement produced by an AI model, such as a wrong address, an invented service or a business that does not exist. It happens because models generate plausible text rather than retrieving verified facts. Clear, consistent information on your website and across reputable directories gives models less room to guess wrongly about you.

The acronyms

The alphabet soup of the field, unpacked one at a time.

SEO (search engine optimisation)

SEO, search engine optimisation, is the practice of improving a website so it ranks higher in the ordinary list of search results on Google or Bing. Success is a position and a click. SEO still matters in the AI era, because AI systems lean heavily on pages that already rank well and are already trusted.

AEO (answer engine optimisation)

AEO, answer engine optimisation, is the work of getting your business named in direct answers: featured snippets, voice replies and AI chat responses. There is no ranked list to appear in, so you are either in the answer or absent from it. Our comparison of AEO, SEO and GEO explains how the three relate.

GEO (generative engine optimisation)

GEO, generative engine optimisation, is the work of being cited or recommended when an AI system composes an answer fresh, as ChatGPT, Gemini and Perplexity do. The unit being optimised is less the page and more the fact: what you do, where you do it, and whether independent sources agree. We compare it with ranking work in GEO vs SEO.

RAG (retrieval-augmented generation)

RAG, retrieval-augmented generation, is the technique where an AI system first retrieves relevant documents, usually live web pages, and then writes its answer from them rather than from memory alone. Most AI search products use it. RAG is why current, crawlable pages can appear in AI answers even when they were published after the model's training ended.

E-E-A-T

E-E-A-T stands for experience, expertise, authoritativeness and trustworthiness. It is the framework Google's human quality raters use to judge content, described in Google's Search Quality Rater Guidelines. It is not a direct ranking score, but the signals associated with it, such as named authors, real credentials and accurate information, also make content a stronger candidate for AI citation.

NAP consistency

NAP consistency means your business name, address and phone number are stated identically everywhere they appear: your website, your Google Business Profile, directories and social profiles. Contradictory listings make it harder for any system, human or machine, to state a fact about you with confidence, which makes inconsistent NAP data one of the most common and most fixable AI visibility problems.

The crawlers

The bots that read your site. The distinction that matters most in this section is the first one.

AI crawler

An AI crawler is an automated program run by an AI company that visits web pages and reads their content, either to build training data or to fetch pages while answering a question. Each crawler identifies itself with a user agent name, such as GPTBot or PerplexityBot, which is how rules in a robots.txt file can target one without touching another.

Training crawler vs retrieval crawler

A training crawler collects pages to help build future AI models; a retrieval crawler fetches pages while a live question is being answered. The distinction carries real consequences: blocking a retrieval crawler removes your pages from AI answers now, while blocking a training crawler only keeps them out of future training data and does not affect whether you are cited today.

GPTBot

GPTBot is OpenAI's training crawler. It collects web content that may be used to train future OpenAI models. Blocking GPTBot in robots.txt keeps your pages out of that training data, but it does not remove you from ChatGPT's live search answers, because those are fetched by the separate OAI-SearchBot.

OAI-SearchBot

OAI-SearchBot is the crawler OpenAI uses for ChatGPT's search feature. It finds and fetches pages so ChatGPT can cite them in live answers. Blocking it in robots.txt removes your site from those answers, so most businesses should allow it. Our guide to getting recommended by ChatGPT walks through the practical steps.

ClaudeBot

ClaudeBot is Anthropic's web crawler, used to gather content for training Claude models. Anthropic publishes separate user agents, such as Claude-User, for the fetches Claude makes when a person asks it about a specific page. As with other AI companies, robots.txt rules for ClaudeBot govern training collection rather than every fetch Claude ever makes.

PerplexityBot

PerplexityBot is the crawler Perplexity uses to build the index behind its AI search engine. Blocking it in robots.txt makes your pages unavailable to Perplexity's answers, which cite their sources prominently. Perplexity also uses a separate Perplexity-User agent when it fetches a page because a user asked about it directly.

Google-Extended

Google-Extended is a robots.txt token that controls whether your content may be used to train Google's Gemini models and for some grounding purposes. It is not a crawler itself, and blocking it does not affect your Google Search rankings or your appearance in AI Overviews, which are governed by ordinary Googlebot indexing.

The plumbing

The files and markup that tell machines, in their own language, what your business is.

robots.txt

robots.txt is a small text file at the root of a website that tells crawlers which parts of the site they may read. Rules target crawlers by name, so it is possible to block AI training crawlers while allowing retrieval crawlers, or to block everything by accident. Misconfigured robots.txt files, often set by a plugin or a past developer, are a common and silent cause of AI invisibility.

llms.txt

llms.txt is a proposed convention: a plain-text file at the root of a website that gives AI systems a curated summary of the site and links to its most useful pages. It is not an official standard, and the major AI companies have not confirmed they use it. It costs little to add, but treat any confident claim about its power sceptically.

Structured data

Structured data is machine-readable code added to a web page that states facts plainly, such as business type, address, opening hours and services. It removes guesswork: instead of inferring facts from your prose, a machine reads them directly. On the modern web it is usually written in the JSON-LD format, using the schema.org vocabulary.

JSON-LD

JSON-LD is the format Google recommends for writing structured data. It is a block of JavaScript Object Notation placed inside a script tag, separate from the visible page content, which makes it easier to add and maintain than markup woven through the HTML. The structured data on this page, including this glossary's own term list, is JSON-LD.

schema.org

schema.org is the shared vocabulary used in structured data, maintained by a community project founded by Google, Microsoft, Yahoo and Yandex. It defines types such as LocalBusiness, Dentist, Product and FAQPage, and the properties each can carry. Using its types correctly is what lets different machines agree on what your page is describing.

Entity

An entity is a distinct thing a machine can identify: a business, a person, a place, a product. Search and AI systems try to resolve mentions across the web to the same entity. A business that is clearly one entity everywhere, with the same name, details and identifiers, is easier to state facts about, and so easier to recommend, than one that looks like several.

Knowledge graph

A knowledge graph is a database of entities and the relationships between them, such as Google's Knowledge Graph. When a system is confident about who you are, where you are and what you do, those facts can be served directly into answers. Consistent public information across your site, your profiles and reputable directories is how a small business earns that confidence.

sameAs

sameAs is a schema.org property that lists other web pages referring to the same entity, typically your Google Business Profile, social profiles and directory listings. It helps machines connect your scattered presences into a single identity. A correct sameAs list in your homepage's structured data is a cheap, useful identity signal that many small business sites simply omit.

The measurements

How being in the answer, or missing from it, actually gets counted.

Grounding

Grounding is the process of tying an AI answer to real, checkable sources rather than the model's memory alone. A grounded answer is composed from retrieved documents and usually cites them. For a business, being retrievable and quotable is what makes you available at grounding time, which is when the answer is actually assembled.

Citation

A citation is a visible reference in an AI answer pointing to the source of a claim, usually as a link. Perplexity, ChatGPT search and AI Overviews all show them. Citations are the clearest evidence that a system read and used a page, which makes them a core unit of AI visibility measurement.

Zero-click search

A zero-click search is one that ends without the searcher visiting any website, because the answer appeared on the results page or in the chat itself. AI answers push more searches toward zero clicks, which means website traffic understates your real presence. You can be influencing customers inside answers your analytics never see.

Share of voice

Share of voice, in AI visibility, is the percentage of relevant AI answers that name your business rather than a competitor. It is measured by asking assistants a set of customer questions on a schedule and recording who gets named. Because AI answers vary between runs, share of voice is read as a trend over time, not as a single result.

Where to start

Definitions are the map, not the territory. If you want to know what AI actually says about your business today, a free scan asks the assistants the questions your customers ask and shows you whether you are named, who is named instead, and what is missing. From there, an audit is $495 as a one-off and ongoing monitoring is $199 a month, in Australian dollars.