Glossary

The vocabulary of GEO.

207 reference terms, tagged Schema.org DefinedTerm for citation by ChatGPT, Claude, Perplexity, Gemini and Copilot. Click a term for its entry.

GEO concepts

AEO (Answer Engine Optimization): AEO is optimisation for engines that answer questions directly instead of listing links.
AI citation: An AI citation is the event of an LLM mentioning your brand in an answer, with or without a link.
AI crawler: An AI crawler is a bot that scans the web to feed an LLM, either for training or real-time retrieval.
AI indexing: AI indexing is the process by which an LLM or answer engine stores and organises a site's pages for retrieval.
AI information retrieval: AI information retrieval is the stage where a hybrid system pulls relevant passages before generating the answer.
AI Overviews (Google): AI Overviews are Google's AI-generated answers displayed above the classic search results.
AI Search: AI Search is the new generation of search engines that answer in natural language with citations.
AI visibility / brand visibility: AI visibility measures how often your brand appears in LLM answers across your target queries.
Answer engine: An answer engine is a search system that answers a question rather than returning a list of links.
Brand-as-entity strategy: Brand-as-entity strategy turns your brand into an entity recognised by search engines and AIs, not just a keyword.
Brand citability: Citability is a brand's propensity to be cited by LLMs. A blend of authority, structure, freshness and clarity.
Brand knowledge graph: A knowledge graph is your brand's structured 'identity card' in search and AI memory.
Brand mention: A brand mention is any occurrence of your brand name in content, whether or not it links back to you.
Brand SERP: The Brand SERP is what Google shows when someone searches your brand name, the 'home page' that LLMs also consult.
Citation rate: Citation rate is the percentage of target queries where your brand is cited by a given LLM.
Conversational query: A conversational query is a question asked in full natural language, the way you would speak to a human.
E-E-A-T: E-E-A-T (Experience, Expertise, Authoritativeness, Trustworthiness) is Google's grid for assessing content and author quality.
Entity co-occurrence: Entity co-occurrence is the fact that your brand appears next to other strong entities (competitors, brands, places, experts).
Entity SEO: Entity SEO treats your brand as an entity (concept, organisation, person) rather than a list of keywords.
Featured snippet vs AI answer: A featured snippet is Google's 'position 0' box (a short extract from a site). An AI answer is generated, longer and multi-source.
Generative answer: A generative answer is an answer produced on the fly by an LLM, distinct from a link or featured snippet.
GEA (Generative Engine Advertising): GEA is the preparation, activation and management of a brand's paid visibility inside generative engines and AI interfaces, wherever those environments offer advertising formats (Google AI Overviews and AI Mode, Copilot Sponsored Answers, upcoming conversational formats). An emerging market, it complements organic GEO: a diagnostic of the platforms available, strategy and budget split, activation on solutions actually on the market, measurement and attribution.
GEO (Generative Engine Optimization): GEO is the discipline of making a brand visible and citable by generative AIs (ChatGPT, Claude, Perplexity, Gemini).
GEO score: The GEO score boils a brand's performance against generative AIs down to a single number.
Knowledge anchor: A knowledge anchor is a fixed authority point (Wikipedia, Wikidata, top-tier press) that anchors your brand in LLMs' memory.
LLM share of voice: LLM share of voice is your brand's share of AI citations in a sector, against competitors.
LLMO (LLM Optimization): LLMO is another term for GEO, popularised by some US consultants.
Long tail query: A long-tail query is a long, precise query with low volume but high intent.
Search Generative Experience (SGE): SGE is Google's internal name for AI Overviews, the feature that answers directly at the top of the results page.
Semantic search: Semantic search understands the intent of a query, not just the words used.
SEO vs GEO: SEO optimises for Google. GEO optimises for generative AIs. Techniques overlap by ~40% and diverge by ~60%.
Source ranking: Source ranking is the order in which an LLM sorts its sources before generating an answer.
Topical authority: Topical authority is the perceived legitimacy of a site on a given topic, measured by content depth and consistency.
Trust signals: Trust signals are clues that a site is reliable, named author, HTTPS, legal pages, press, years in business.
Vector search: Vector search is search based on semantic proximity between embeddings, not exact words.
Zero-click search: A zero-click search is one where the user gets the answer directly in Google or an LLM, without clicking any link.

AI & LLMs

AI agent / agentic AI: An AI agent is an LLM capable of taking actions, booking a ticket, writing an email, browsing the web, not just answering.
AI hallucination: A hallucination is a confident-sounding but false LLM answer, an invented fact, a fabricated quote or a wrong number.
ChatGPT Enterprise / Claude Enterprise: Enterprise versions of ChatGPT and Claude with confidentiality guarantees (your data is not used for training) and SOC 2 compliance.
ChatGPT (OpenAI): ChatGPT is OpenAI's AI assistant, launched in November 2022. It is the most-used consumer LLM, with over 700 million weekly users in 2026.
Claude (Anthropic): Claude is Anthropic's AI assistant, a direct competitor to ChatGPT. It is widely praised for long-form writing and nuanced reasoning.
Context window: The context window is the maximum amount of text an LLM can process in a single conversation. The wider it is, the more the model can 'see' at once.
Copilot (Microsoft): Copilot is Microsoft's AI assistant, integrated into Bing, Windows, Edge, Office 365 and GitHub.
Deep Research mode: Deep Research is an advanced mode (ChatGPT, Perplexity, Gemini) in which the AI runs a multi-source investigation for several minutes and produces a structured report.
DeepSeek / Qwen (Chinese models): DeepSeek and Qwen are the main Chinese LLMs, often open source and highly performant at a low cost.
Embedding / semantic vector: An embedding turns a piece of text into a list of numbers (a vector) that captures its meaning. Similar texts have similar embeddings.
Explicit citation vs absorption: An LLM can cite a source explicitly (with a link) or simply 'absorb' your knowledge without naming you.
Fine-tuning: Fine-tuning means re-training an existing LLM on your own data so it learns your style or domain.
Foundation model: A foundation model is a general-purpose, pre-trained LLM that can be repurposed across many tasks.
Gemini (Google): Gemini is Google's generative AI, integrated into Search through AI Overviews and across the entire Workspace suite.
Generative AI: Generative AI is artificial intelligence that produces original content, text, image, audio or video, instead of merely classifying or predicting.
Hybrid AI system (LLM + web search): A hybrid system pairs an LLM with real-time web search, Perplexity, ChatGPT Search, Gemini with AI Overviews.
Inference: Inference is the act of running an LLM to produce an answer, as opposed to training it.
Jailbreak prompt: A jailbreak is a prompt that bypasses an LLM's safety rules to make it produce forbidden content.
Knowledge cut-off: The knowledge cut-off is the date beyond which an LLM has no knowledge. GPT-4o is cut off at October 2023, Claude Opus 4 at January 2026.
LLM API: An LLM API lets a developer use an LLM inside their own software, on a pay-as-you-go basis.
LLM (Large Language Model): An LLM is an AI trained to understand and generate text. ChatGPT, Claude and Gemini are consumer-facing LLMs.
LLM latency: Latency is the time between your question and the LLM's complete answer. Measured in seconds, it shapes how the tool can be used.
LLM tech stack 2026: The LLM tech stack is the set of tools combined to build an AI assistant in 2026, model, vector store, framework, monitoring.
Loss function: The loss function measures an LLM's error during training. The lower it gets, the better the model has learned.
Mistral / Le Chat: Mistral AI is Europe's flagship generative-AI company, based in Paris. Its consumer assistant is called Le Chat.
Multimodality: Multimodality is the ability of an AI to process several content types, text, image, sound, video, simultaneously.
Open source vs closed source LLMs: Open-source LLMs (Llama, Mistral, Qwen) ship public weights. Closed-source LLMs (GPT, Claude, Gemini) are proprietary.
Perplexity: Perplexity is an AI search engine that answers questions directly with cited sources. A clear alternative to Google Search.
Plugin / custom GPT: A custom GPT is a mini-assistant created by a ChatGPT user, specialised on a use case (a profession, a brand, a topic).
Pre-training vs post-training: Pre-training teaches a model the language on terabytes of text. Post-training makes it useful and safe.
Prompt / prompting: A prompt is the instruction you give an LLM. Prompting is the craft of phrasing that instruction to get the best answer.
RAG (Retrieval Augmented Generation): RAG is a technique that pairs an LLM with a document store. The LLM searches first, then answers while citing its sources.
Reasoning model (o1, o3): Reasoning models (OpenAI's o1, o3; Claude Opus 4 thinking mode) 'think' before answering. Slower, but more accurate.
Safety alignment / guardrails: Safety alignment of an LLM is the set of internal rules that stop it from producing dangerous, illegal or defamatory content.
Sora / Veo (generative video): Sora (OpenAI) and Veo (Google) are the leading AI video-generation models, from a few seconds to several minutes from a text prompt.
Synthetic answer / answer engine: A synthetic answer is a complete AI-generated answer that replaces the list of links of a classic search engine.
Temperature (LLM parameter): Temperature controls how creative an LLM is. At 0 the model is highly predictable; at 1+ it gets inventive and sometimes unpredictable.
Token cost / API pricing: The cost of using an LLM via API is billed per token, input tokens plus output tokens.
Token (LLM text unit): A token is the basic unit an LLM processes, roughly 0.75 words in English. 1,000 tokens ≈ 750 words.
Top-K, top-P sampling: Top-K and top-P are technical parameters that control how varied an LLM's answers are.

Technical standards

200 vs 301 vs 410 status codes: HTTP status codes tell a bot the state of a page: 200 = OK, 301 = permanent redirect, 410 = permanently deleted.
ai-content manifest: An ai-content manifest is a file declaring the AI nature of content (generated, augmented, human-verified).
ai.txt (variant): ai.txt is a variant of llms.txt proposed by Spawning, focused on opt-out from AI training for creators.
Applebot vs Applebot-Extended: Applebot indexes for Siri and Spotlight. Applebot-Extended specifically targets training of Apple Intelligence models.
Bing Webmaster Tools AI Performance: The AI Performance report in Bing Webmaster Tools shows how your site appears in Bing Copilot and AI answers.
Canonical URL: The canonical URL is the 'official' version of a page when multiple URLs serve the same content.
CCBot (Common Crawl): CCBot is the crawler of Common Crawl, a non-profit foundation that scrapes the web and publishes a dataset used by nearly every LLM.
ClaudeBot: ClaudeBot is Anthropic's crawler that fetches content to train and feed Claude.
Cloudflare AI Crawl Control: Cloudflare AI Crawl Control is a dashboard letting you view, allow or block AI crawlers in one click.
CLS (Cumulative Layout Shift): CLS measures unexpected layout shifts during loading, when a button 'jumps' just as you try to click it.
Core Web Vitals: Core Web Vitals are Google's three metrics measuring a page's user experience: speed, responsiveness, visual stability.
Crawler budget: Crawler budget is the number of pages a bot is willing to crawl on your site within a given time window.
Google-Extended: Google-Extended is the robots.txt token that controls how Google's AI models (Gemini, Vertex AI) use your content.
GPTBot: GPTBot is OpenAI's crawler that collects web content to train future GPT models.
hreflang: hreflang is the HTML attribute that tells engines which language or country a page targets.
INP (Interaction to Next Paint): INP measures the time between a user action (click, tap) and the page's next visual refresh.
JavaScript indexability: JavaScript indexability is a bot's ability to execute a page's JS to 'see' its real content.
JSON Feed: JSON Feed is a modern JSON alternative to RSS/Atom feeds for syndicating content.
JSON-LD: JSON-LD is the technical format Google recommends for embedding structured data (schema.org) in a page.
LCP (Largest Contentful Paint): LCP is the time it takes to display the largest visible element of a page (hero image, headline).
Lighthouse audit: Lighthouse is Google's open-source tool that audits performance, accessibility, SEO and best practices on a page.
LLM API webhooks: A webhook is a URL where an AI system can send a notification or event in real time.
llms-full.txt: llms-full.txt is the extended version of llms.txt: the full site content condensed for LLMs.
llms.txt: llms.txt is a text file at the site root that tells AIs which content to prioritise and how to navigate.
Microdata vs RDFa: Microdata and RDFa are older formats for embedding schema.org in HTML, less preferred than JSON-LD.
MistralBot: MistralBot is Mistral AI's crawler that powers Le Chat and the European sovereign models.
Mobile-first indexing: Mobile-first indexing means Google indexes the mobile version of your site first. Desktop is secondary.
Modern robots.txt: robots.txt is the file that tells bots which paths to crawl. The 2026 take distinguishes training and search bots.
OAI-SearchBot: OAI-SearchBot is OpenAI's crawler that fetches content for ChatGPT Search in real time (distinct from GPTBot).
Open Graph: Open Graph is Facebook's meta-tag standard controlling how your page previews on social networks.
Opt-in vs opt-out crawlers: Opt-in and opt-out describe who decides a site's inclusion in an AI crawler's dataset.
PageSpeed Insights: PageSpeed Insights is Google's free tool measuring a page's performance on mobile and desktop.
Pay-Per-Crawl (Cloudflare): Pay-Per-Crawl is a Cloudflare mechanism that lets a publisher charge AI bots for each page crawled.
PerplexityBot: PerplexityBot is the crawler of Perplexity AI feeding its index in real time.
RSS / Atom: RSS and Atom are the historical content-syndication formats, used since 1999 to track sites.
Schema.org: Schema.org is the universal vocabulary for describing a page's content in a structured way, readable by engines and AIs.
Sitemap index: A sitemap index is a file listing several sitemap.xml files, useful for large sites (>50,000 URLs).
Sitemap.xml: sitemap.xml is a file listing all important URLs on your site, helping bots find them.
SSR (Server-Side Rendering): SSR renders a page's HTML on the server, as opposed to CSR (Client-Side Rendering) which depends on browser JavaScript.
Twitter Cards: Twitter Cards (X Cards) are the meta-tags that control how your page previews on X (formerly Twitter).

Schema.org

JSON-LD @graph: @graph is the JSON-LD property that lets you bundle several schemas in one coherent block.
sameAs (Wikidata, LinkedIn): sameAs is the schema property linking an entity to its official pages elsewhere on the web, Wikidata, LinkedIn, Wikipedia.
Schema.org Article: The Article schema describes editorial content (article, blog post, news) for engines and AIs.
Schema.org Author: The author property (of type Person or Organization) attributes content to its creator.
Schema.org Brand: The Brand schema identifies a commercial brand as an entity distinct from a company.
Schema.org BreadcrumbList: The BreadcrumbList schema describes the breadcrumb of a page, Home > Category > Page.
Schema.org ContactPage: The ContactPage schema is the WebPage subtype specific to your contact page.
Schema.org Course: The Course schema describes a training programme, online course or academic curriculum.
Schema.org DefinedTerm: The DefinedTerm schema marks up each glossary entry to make it usable by engines and AIs.
Schema.org Event: The Event schema describes a physical or virtual event, concert, conference, fair, webinar.
Schema.org FAQPage: The FAQPage schema marks up a Q&A page so it can be used in featured snippets and AI answers.
Schema.org Hotel / LodgingBusiness: The Hotel and LodgingBusiness schemas describe accommodation, rooms, stars, services.
Schema.org HowTo: The HowTo schema describes a step-by-step procedure, recipe, instructions, tutorial.
Schema.org JobPosting: The JobPosting schema describes a job offer, title, employer, location, salary, conditions.
Schema.org LocalBusiness: The LocalBusiness schema describes a local business, address, hours, contact, geolocation.
Schema.org MedicalProcedure: The MedicalProcedure schema describes a medical procedure, diagnosis, treatment, intervention.
Schema.org OfferCatalog: The OfferCatalog schema lists several grouped offers or services, product line, menu, service slate.
Schema.org Organization: The Organization schema describes a company, its contacts, executives and official identifiers.
Schema.org Person: The Person schema describes a real person, executive, author, expert, as an identifiable entity.
Schema.org Product: The Product schema describes a sold item, name, brand, price, image, availability.
Schema.org Recipe: The Recipe schema describes a cooking recipe, ingredients, steps, time, calories.
Schema.org Restaurant: The Restaurant schema identifies a dining establishment, address, opening hours, cuisine type, menu.
Schema.org Review / AggregateRating: The Review and AggregateRating schemas describe reviews and average ratings for a product, place or service.
Schema.org SearchAction: The SearchAction schema exposes your internal search engine so it can appear in Google's sitelinks searchbox.
Schema.org Service: The Service schema describes a commercial service, coaching, audit, consulting, transport.
Schema.org SoftwareApplication: The SoftwareApplication schema describes a software product or app, features, OS, price, ratings.
Schema.org SpecialAnnouncement: The SpecialAnnouncement schema flags an urgent announcement, closure, alert, COVID, extraordinary event.
Schema.org validator: The Schema.org validator (Google Rich Results Test, validator.schema.org) verifies that your markup is correct.
Schema.org WebPage: The WebPage schema is the generic type for any web page, usually refined by a specific subtype.
Schema.org WebSite: The WebSite schema identifies the site itself as an entity, distinct from individual pages.

Measurement & KPIs

AI click-through rate: AI CTR is the percentage of users who click your link from an AI answer.
AI traffic referrer: AI traffic referrer is the web traffic that lands on your site via a click from an AI answer.
Bot crawl frequency: Bot crawl frequency is the number of times a bot visits your site over a period.
Brand authority score: Brand authority score sums up your brand's overall authority (entity, press, backlinks, anchors).
Brand mention frequency: Brand mention frequency is the number of times your name appears on the web over a given period.
Citation depth: Citation depth measures the length of the passage where your brand is cited, single name, short paragraph, developed paragraph.
Citation drift: Citation drift is the over-time evolution of how LLMs cite your brand, phrasing, context, position.
Citation footprint: Citation footprint is the total imprint of your brand across AI citations, all LLMs combined.
Citation overlap (multi-LLM): Citation overlap measures whether your brand is cited by several LLMs in parallel for the same query.
Citation weighting: Citation weighting assigns a different weight to each citation based on quality (LLM, position, context).
Knowledge graph coverage: Knowledge graph coverage signals whether your brand exists in public knowledge graphs (Wikidata, Google KG, Wikipedia).
LLM share of voice: LLM share of voice measures your brand's share of AI citations in a sector versus competitors.
Mention sentiment: Mention sentiment indicates whether an AI citation is positive, neutral or negative toward you.
Multilingual coverage: Multilingual coverage is the share of target languages where your brand is cited.
Position in answer: Position-in-answer indicates where your brand appears in an AI answer, first named, middle, end of list.
Quality signals score: Quality signals score aggregates the technical and editorial signals that make a site 'good' in engines' eyes.
Reverse citation: Reverse citation is the event of an LLM citing ONE of your external sources to talk about you.
Sectoral coverage rate: Sectoral coverage rate is the percentage of sector queries where your brand is cited at least once.
Time-to-citation (T+Days): Time-to-citation is the delay between publishing content and its first citation by an LLM.
Topical relevance score: Topical relevance score measures how strongly your site is perceived as expert on a specific topic.

Players & platforms

Aleyda Solis (Orainti): Aleyda Solis is a Spanish SEO/GEO consultant, founder of Orainti and a global figure in the field.
Anthropic: Anthropic is the US firm behind Claude, the leading ethical alternative to ChatGPT, founded by former OpenAI staff.
AthenaHQ: AthenaHQ is another US GEO measurement tool, a Profound competitor, launched in 2024.
Bing Webmaster Tools: Bing Webmaster Tools is Microsoft's equivalent of Google Search Console, free and under-used in Europe.
BrightEdge: BrightEdge is a legacy enterprise SEO platform that added a GEO module in 2024.
ChatGPT plugins / GPTs: ChatGPT plugins (rebranded GPTs in 2023) are specialised assistants built by users or companies.
Cloudflare: Cloudflare is the leading CDN/security operator of the web, and the de-facto arbiter of the AI crawler market since 2024.
Cohere: Cohere is a Canadian generative-AI startup specialised in enterprise and multilingual use cases.
Conductor: Conductor is a US enterprise SEO platform, peer of BrightEdge, that also pivoted to GEO in 2024.
Google DeepMind: Google DeepMind is Google's AI division, merging Google Brain and DeepMind (London) in 2023.
Inflection AI: Inflection AI was a consumer AI startup (Pi), largely absorbed by Microsoft in 2024.
iPullRank (Mike King): iPullRank is Mike King's US SEO/GEO agency, one of the recognised pioneers and theorists of GEO.
Lily Ray (Algorythmic): Lily Ray is a US SEO/GEO consultant (ex-Path Interactive, now Algorythmic), specialised in E-E-A-T and YMYL.
Meta AI (Llama): Meta AI is the AI division of Meta (Facebook/Instagram/WhatsApp). Publishes the Llama models as free open-weight releases.
Microsoft / Azure OpenAI: Microsoft is one of the main enterprise AI distributors via Copilot and the Azure OpenAI Service.
Mistral AI: Mistral AI is Europe's flagship generative-AI company, based in Paris.
OpenAI: OpenAI is the US company behind ChatGPT, GPT-4, DALL-E and Sora. The global pioneer of consumer generative AI.
Perplexity AI: Perplexity AI is the US startup behind the Perplexity AI search engine, an open alternative to Google Search.
Proema Insight: Proema Insight is a European GEO measurement tool tracking AI citations for B2B brands.
Profound: Profound is a US tool that measures a brand's AI citations on ChatGPT, Perplexity, Claude and Gemini.
Schema.org Foundation: The Schema.org Foundation is the body maintaining the schema.org vocabulary, originally Google, Bing, Yahoo and Yandex.
Search Engine Journal: Search Engine Journal (SEJ) is one of the leading English-language publications on SEO and GEO.
Search Engine Land: Search Engine Land is the other major English-language SEO/SEM/GEO publication, complementary to SEJ.
Wikipedia / Wikidata: Wikipedia is the online encyclopedia; Wikidata is its structured database. They are the two most-cited knowledge anchors by LLMs.
xAI (Grok): xAI is Elon Musk's AI company, behind Grok, integrated into X (Twitter).

Regulation

AI whistleblower: An AI whistleblower is a person who reports abusive or dangerous AI use within an organisation.
APD (Belgium): The APD (Data Protection Authority) is the Belgian counterpart to France's CNIL.
Article 50 EU AI Act: Article 50 requires any organisation to disclose that content was generated or modified by AI.
Article 53 EU AI Act: Article 53 regulates foundation models (GPT, Claude, Gemini, Mistral), documentation and transparency duties.
Bartz v. Anthropic: Andrea Bartz, an author, sued Anthropic in 2024 over the use of her books to train Claude.
CCPA (California): CCPA is California's personal-data protection law, a partial US counterpart to GDPR.
CNIL (France): The CNIL is France's data protection authority, the French counterpart of Belgium's APD.
Copyright and AI: Copyright protects original works, text, image, code. Its application to generative AI is being shaped by ongoing 2026 case law.
Data Privacy Framework (EU-US): The Data Privacy Framework is the EU-US agreement enabling lawful transfer of personal data between Europe and the United States.
EDPB: The EDPB (European Data Protection Board) is the EU body coordinating national data protection authorities.
EU AI Act: The EU AI Act is the first comprehensive European AI regulation, in force since 2024.
GDPR: GDPR is the European personal-data protection regulation, in force since 2018.
NYT vs OpenAI case: The New York Times sued OpenAI and Microsoft in December 2023 for unauthorised use of its articles to train GPT.
Opt-in / opt-out crawlers: The opt-in vs opt-out debate is about the default: is your content included or excluded unless you say otherwise?
Registered vs well-known trademark: A registered trademark is officially registered (EUIPO, INPI, BOIP). A well-known trademark is recognised through use, even without registration.
Unfair competition (BE/FR): Unfair competition covers misleading, disparaging or parasitic commercial practices between businesses.

Get started

Is your brand citable by AI?

Receive your free GEO score and then request a full GEO diagnostic: what ChatGPT, Perplexity and Gemini do (or don’t) see about you today.

Request a diagnostic See the method