AI Search Glossary - Every Term Defined

AI search optimisation has its own vocabulary. This glossary defines every key term - from AEO and GEO to Speakable schema and crawl-to-referral ratios - with cross-references to our detailed guides and to primary sources such as Schema.org, Google Search Central, and the platform documentation itself. Bookmark this page as your reference for AI search terminology.

A

AEO (Answer Engine Optimisation)

AEO is the practice of optimising content to appear as direct answers in search engine features like featured snippets, voice assistant responses, and People Also Ask boxes. It predates the generative AI era and is narrower in scope than GEO. AEO focuses specifically on being the selected answer, not on appearing in synthesised multi-source responses.

See: AEO vs GEO vs LLMO comparison | Reference: Google Search Central on featured snippets

AI Mode

AI Mode is Google's dedicated conversational search interface within Google Search, powered by Gemini. It allows users to ask follow-up questions and have multi-turn conversations while receiving AI-generated answers with cited sources. AI Mode represents Google's response to ChatGPT and Perplexity.

Reference: Google Search product announcements

AI Overviews

AI Overviews are AI-generated summary answers that appear at the top of some Google search results. They draw on pages that already rank in Google's organic results, and Google has stated that there is no separate way to mark up content for AI Overviews beyond standard search best practices. AI Overviews were previously launched under the name Search Generative Experience (SGE).

See: How to get cited in AI Overviews | Reference: Google Search Central: AI features and your website

Answer Capsule

An answer capsule is a 40-60 word factual paragraph placed immediately after a heading, designed to be extracted by AI platforms as a citation-ready response. Answer capsules contain specific data points, direct answers, and verifiable claims. When combined with Speakable schema, they become the primary extraction target for AI voice and text responses.

See: How to structure your website for AI crawlers

areaServed

areaServed is a schema.org property used in Organisation schema to specify the geographic area a business serves. AI platforms use areaServed to match businesses with location-specific queries. It accepts Country, State, City, or GeoShape values and is particularly important for local and regional businesses seeking AI visibility.

Reference: schema.org/areaServed

Article Schema

Article schema is a schema.org type that marks up news articles, blog posts, and editorial pages with properties such as headline, author, datePublished, and publisher. It helps search engines and AI platforms identify the author, the publishing organisation, and the freshness of a page, all of which feed E-E-A-T and citation decisions.

See: Schema markup for AI search | Reference: schema.org/Article, Google Article documentation

B

Bing Webmaster Tools

Bing Webmaster Tools is Microsoft's free platform for submitting sitemaps, inspecting URLs, and monitoring how Bing crawls and indexes a site. Because ChatGPT and Microsoft Copilot draw on Bing's index, getting indexed in Bing is a prerequisite for appearing in those AI answers. It also supports the IndexNow protocol for instant submission.

See: Bing submission guide | Reference: Bing Webmaster Tools

BreadcrumbList

BreadcrumbList is a schema.org type that defines the navigational hierarchy of a page within a website. AI crawlers use BreadcrumbList to understand site structure and the topical relationship between pages. It helps establish which pages are parent topics and which are subtopics, improving content categorisation during AI indexing.

See: Schema markup for AI search | Reference: schema.org/BreadcrumbList, Google breadcrumb documentation

C

ChatGPT

ChatGPT is OpenAI's conversational AI platform and the most widely used AI assistant for search-style queries. Its web search feature retrieves live results via Bing rather than Google. Consistent brand mentions across the web and an established entity presence are widely cited as the strongest factors influencing whether ChatGPT references a source.

See: How to get cited in ChatGPT | Reference: OpenAI bots and crawlers documentation

Citation (AI)

In AI search, a citation is when an AI platform references, quotes, or attributes information to a specific source in its generated answer. Citations may include direct links (Perplexity, ChatGPT) or brand mentions without links (Claude, some Gemini responses). Citation tracking is the AI equivalent of rank tracking in traditional SEO.

ClaudeBot

ClaudeBot is Anthropic's web crawler that collects content used to help train Claude. Like most training crawlers, it accesses far more pages than it sends referral traffic for. You can allow or disallow ClaudeBot in robots.txt to control whether your content is available for Anthropic's crawling.

See: How to get cited in Claude | Reference: Anthropic crawler documentation

Copilot (Microsoft)

Microsoft Copilot is an AI assistant integrated into Bing, Windows, Microsoft 365, and Edge. It uses Bing's search index, making it closely aligned with ChatGPT in terms of content discovery. Optimising for Bing visibility simultaneously improves visibility in both ChatGPT and Copilot.

Reference: Bing Webmaster Tools help

Crawl-to-Referral Ratio

The crawl-to-referral ratio measures how many pages an AI platform crawls for every one referral visit it sends back. Search-and-citation crawlers (such as those behind Perplexity) tend to have lower ratios, while training crawlers access many pages but send little or no referral traffic. A lower ratio means the platform is more likely to send traffic to sources it cites.

See: Platform comparison with ratio data

E

Embeddings

An embedding is a numerical vector representation of text that captures its meaning, allowing AI systems to measure how semantically similar two pieces of content are. Retrieval-augmented generation relies on embeddings to find the most relevant passages for a query. Clear, self-contained passages tend to produce cleaner embeddings and are easier for AI systems to retrieve and cite.

Reference: Word embedding (Wikipedia)

E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness)

E-E-A-T is Google's content quality framework that evaluates Experience, Expertise, Authoritativeness, and Trustworthiness. AI platforms - particularly Gemini and AI Overviews - inherit these signals from Google's index. Demonstrating E-E-A-T through author credentials, original data, and authoritative backlinks improves AI citation likelihood.

Reference: Google: creating helpful, reliable, people-first content, SEO (Wikipedia)

Entity SEO

Entity SEO is the practice of building a consistent, verifiable digital identity for a brand, person, or concept across multiple platforms. AI platforms cross-reference entity data from Organisation schema, Google Business Profile, Wikidata, LinkedIn, and other sources. Entity consistency is a prerequisite for AI visibility, particularly in Gemini.

See: How Gemini uses entity relationships | Reference: Knowledge Graph (Wikipedia)

F

Fine-Tuning

Fine-tuning is the process of further training a pre-trained large language model on a narrower dataset to specialise its behaviour. It is distinct from retrieval-augmented generation, which supplies information at query time without changing the model. Most public AI search platforms cite the live web through retrieval rather than fine-tuning, so getting cited depends on retrievability, not on being in the training set.

Reference: Fine-tuning (Wikipedia)

FAQPage Schema

FAQPage is a schema.org type that marks up question-and-answer content. Google deprecated FAQ rich results for most sites in 2023 and no longer shows FAQ snippets from the majority of pages, so FAQPage markup now delivers little to no search-appearance benefit. SEOCompare does not use FAQPage schema. Well-structured question-and-answer content in plain HTML remains valuable for AI extraction without the markup.

See: Schema markup for AI search | Reference: schema.org/FAQPage, Google: changes to FAQ and HowTo rich results

G

Gemini

Gemini is Google's family of AI models, which powers AI Overviews and AI Mode and draws on Google's search index and Knowledge Graph. Gemini weighs entity relationships heavily, so strong traditional SEO and a consistent entity footprint are effectively prerequisites for Gemini visibility.

See: How to get cited in Gemini | Reference: Google DeepMind: Gemini

GEO (Generative Engine Optimisation)

GEO (Generative Engine Optimisation) is the broadest and most commonly used term for AI search optimisation. It covers getting your content cited in AI-generated answers across all platforms - ChatGPT, Gemini, Perplexity, AI Overviews, Copilot, and Claude. GEO encompasses both traditional AEO tactics and newer AI-specific strategies like entity SEO and multi-index optimisation. The term was introduced in a 2023 research paper by Aggarwal et al.

See: AEO vs GEO vs LLMO comparison | Best GEO agencies | Reference: GEO: Generative Engine Optimization (arXiv)

Google Business Profile

Google Business Profile (formerly Google My Business) is Google's free listing for businesses, covering name, address, phone, category, hours, and reviews. It is a primary entity source for Google's Knowledge Graph, so a complete and consistent profile strengthens entity verification used by Gemini and AI Overviews, particularly for local and service-area businesses.

Reference: Google Business Profile

Google Search Console

Google Search Console is Google's free tool for monitoring how a site performs in Google Search, including impressions, clicks, indexing status, and structured-data validity. It is the primary way to confirm that pages are indexed and eligible to appear in Google results and, by extension, in AI Overviews, which draw on the same index.

Reference: Google Search Console

GPTBot

GPTBot is OpenAI's web crawler used for collecting training data for GPT models. It is separate from OAI-SearchBot (which handles real-time ChatGPT search queries). Allowing GPTBot in robots.txt permits your content to be included in future model training. Blocking GPTBot does not prevent ChatGPT from citing you via Bing - OAI-SearchBot handles that.

Reference: OpenAI bots and crawlers documentation

H

Hallucination

A hallucination is when an AI model generates information that is plausible-sounding but false or unsupported by any real source. Hallucinations are a key reason AI platforms favour retrievable, well-structured, verifiable content - clear citations and consistent entity data reduce the chance a model invents or misattributes details about your brand.

Reference: Hallucination in AI (Wikipedia)

hreflang

hreflang is an HTML attribute and HTTP signal that tells search engines which language and regional version of a page to serve. Correct hreflang prevents the wrong-language page being indexed or cited, which matters for UK businesses that want AI platforms to surface their UK content rather than a US or generic variant.

Reference: Google: localised versions of pages

I

IndexNow

IndexNow is a protocol that notifies search engines instantly when content is published or updated. Supported by Bing, Yandex, Seznam, and Naver (not Google). Since ChatGPT uses Bing's index, IndexNow provides near-instant content availability in ChatGPT search results, rather than waiting for Bing's normal crawl cycle.

See: Website structure for AI crawlers | Reference: IndexNow protocol

Information Gain

Information gain is a concept where content provides unique data, insights, or perspectives not available elsewhere on the web. AI platforms favour content with high information gain because it adds to their knowledge base rather than duplicating existing information. Original research, proprietary data, and unique expert analysis all contribute to information gain.

J

JSON-LD

JSON-LD (JavaScript Object Notation for Linked Data) is the format Google recommends for adding structured data to a page. It places schema.org markup in a script block separate from the visible HTML, making it the cleanest way to declare Article, Organisation, Person, and other entity data that AI platforms parse.

See: Schema markup for AI search | Reference: Google: intro to structured data, json-ld.org

K

Knowledge Graph

Google's Knowledge Graph is a database of entities (people, places, organisations, concepts) and the relationships between them. Gemini and AI Overviews use the Knowledge Graph to verify entity claims and understand relationships. Having your brand in the Knowledge Graph - via Wikidata, Wikipedia, or Google Business Profile - significantly improves AI visibility in Google's ecosystem.

Reference: Knowledge Graph (Wikipedia)

L

LLMO (Large Language Model Optimisation)

LLMO is the most technical approach to AI search optimisation, focused specifically on how large language models absorb and retrieve information. It covers training data inclusion, RAG (retrieval-augmented generation), embedding optimisation, and model-specific behaviour differences. LLMO is most relevant for enterprise brands and technical products.

See: AEO vs GEO vs LLMO comparison | Reference: Large language model (Wikipedia)

llms.txt

llms.txt is a proposed standard file placed at a website's root (like robots.txt) intended to give AI models a machine-readable index of the site's most important content. Adoption is early and, as of 2026, no major AI platform has publicly confirmed it uses llms.txt for ranking or retrieval, so it should be treated as experimental.

See: Website structure for AI crawlers | Reference: llmstxt.org proposal

O

OAI-SearchBot

OAI-SearchBot is OpenAI's crawler specifically for real-time ChatGPT search queries, separate from GPTBot (which handles training data collection). When a ChatGPT user asks a question requiring current information, OAI-SearchBot fetches and surfaces relevant pages. Blocking OAI-SearchBot prevents your content from appearing in ChatGPT's real-time search answers.

Reference: OpenAI bots and crawlers documentation

Organisation Schema

Organisation schema is a schema.org type that defines a business entity with properties including @id, name, sameAs, areaServed, and foundingDate. The @id property creates a persistent identifier that other schema blocks can reference. The sameAs property links to authoritative profiles (LinkedIn, Wikidata, GBP) that AI platforms use for entity verification.

See: Schema markup for AI search | Reference: schema.org/Organization

P

Perplexity

Perplexity is an AI answer engine that operates in real time, citing its sources inline with links, and tends to send proportionally more referral traffic than training-focused crawlers. It draws heavily on community and forum content alongside primary sources, and runs a Publishers' Programme offering revenue sharing and citation analytics.

See: How to get cited in Perplexity | Perplexity Publishers' Programme | Reference: Perplexity crawler documentation

Person Schema

Person schema is a schema.org type that defines an individual with properties including name, jobTitle, worksFor, sameAs, and url. It connects content to a credentialed, identifiable author, which AI platforms and search engines use as an expertise and trust signal under E-E-A-T. Linking Article schema to a populated Person entity strengthens author attribution.

See: Schema markup for AI search | Reference: schema.org/Person

Q

Query Fan-Out

Query fan-out is the process where AI platforms break a complex user query into multiple sub-queries, each answered independently before synthesising a combined response. AI Overviews and Gemini use query fan-out extensively. This means your content can be cited for a sub-query even if it doesn't directly match the original complex question.

R

robots.txt

robots.txt is a file at a website's root that tells crawlers which paths they may or may not access. It is the standard mechanism for allowing or blocking AI crawlers such as GPTBot, OAI-SearchBot, ClaudeBot, and Google-Extended. SEOCompare allows all AI crawlers in robots.txt so its content remains eligible for citation across platforms.

See: Website structure for AI crawlers | Reference: Google: robots.txt introduction

RAG (Retrieval-Augmented Generation)

RAG is the technical process where an AI model retrieves relevant documents from an index before generating a response. ChatGPT uses RAG with Bing's index. Perplexity uses RAG with multiple sources. RAG is what allows AI platforms to cite current information rather than relying solely on training data. Optimising for RAG means optimising for the retrieval step - indexation, relevance, and content structure.

Reference: Retrieval-augmented generation (Wikipedia)

S

sameAs

sameAs is a schema.org property that links an entity to its authoritative profiles elsewhere on the web - for example a LinkedIn page, Wikidata item, Companies House record, or Google Business Profile. AI platforms follow sameAs links to confirm that a brand or person is who they claim to be, making it one of the most important properties for entity verification.

See: Schema markup for AI search | Reference: schema.org/sameAs

Schema.org

Schema.org is a collaborative vocabulary for structured data, founded by Google, Microsoft, Yahoo, and Yandex. It defines the types (Organisation, Person, Article, Product) and properties used in JSON-LD markup. Schema.org is the shared dictionary that lets search engines and AI platforms interpret the meaning of page content rather than guessing from the text alone.

See: Schema markup for AI search | Reference: schema.org

SEO (Search Engine Optimisation)

SEO is the practice of improving a website's visibility in organic search results. Because AI Overviews and ChatGPT both rely on existing search indexes (Google and Bing respectively), traditional SEO - indexation, relevance, site quality, and authority - remains the foundation that AI search optimisation is built on rather than a separate discipline.

See: What is AI search optimisation? | Reference: Google SEO Starter Guide, SEO (Wikipedia)

Sitemap (XML)

An XML sitemap is a file listing a site's important URLs to help search engines discover and crawl them. Submitting a sitemap through Google Search Console and Bing Webmaster Tools speeds up indexation, which is the precondition for a page being eligible to appear in AI Overviews or ChatGPT search answers.

Reference: Google: sitemaps overview, sitemaps.org protocol

Structured Data

Structured data is information added to a page in a standardised format (usually schema.org vocabulary expressed as JSON-LD) so machines can understand it unambiguously. It is the umbrella term covering Organisation, Person, Article, and other schema types. Structured data helps AI platforms extract facts, attribute authorship, and verify entities without parsing prose.

See: Schema markup for AI search | Reference: Google structured data gallery

Speakable Schema

Speakable schema (SpeakableSpecification) marks content as suitable for text-to-speech and AI voice assistants. It uses CSS selectors or XPaths to target specific content blocks (like answer capsules) that platforms should prioritise for spoken responses. Google lists Speakable as a limited beta feature for news content, so support is partial rather than universal.

See: Schema markup for AI search | Reference: schema.org/speakable, Google Speakable documentation

SXO (Search Experience Optimisation)

SXO combines traditional SEO with user experience (UX) optimisation, focusing on the complete search journey from query to conversion. In the AI search context, SXO extends to ensuring that users who discover your brand through AI answers have a seamless experience when they visit your site. Page speed, mobile experience, and clear conversion paths all fall under SXO.

T

Topical Authority

Topical authority is the depth and breadth of content a website has on a specific subject. AI platforms favour sources that demonstrate comprehensive topic coverage over sites with isolated articles. Building topical authority means creating clusters of interlinked content that cover a subject from multiple angles, with internal links connecting related pages.

W

Wikidata

Wikidata is a free, structured knowledge base maintained by the Wikimedia Foundation. It provides machine-readable entity data that AI platforms - particularly Gemini - cross-reference for entity verification. Creating a Wikidata entry for your organisation with consistent data (name, description, location, sameAs links) strengthens your entity signal across all AI platforms.

Reference: Wikidata