Skip to content

Wiki

Encyclopedia entries for GEO concepts, terms, standards, companies, people, and events.

Topic: Foundations (10) Signals (10) Infrastructure (10) Practice (1)

Foundations

  • AEO vs GEO
    Concept In practice, AEO and GEO describe the same work. AEO emerged in the era of Featured Snippets and voice assistants; GEO became the prevailing term as generative, multi-source answers became the default.
  • AI Search Timeline (2022–present)
    Concept AI search began with ChatGPT on 2022-11-30 and moved through four phases: Shock (Bing Chat and Bard), Fragmentation (SGE, Claude, Perplexity, and the coinage of GEO), Mainstream (AI Overviews and ChatGPT Search), and Stabilization (AI Mode reaching 1B MAU).
  • Answer Loop
    Concept Every generative answer follows the same four-step runtime loop: query understanding, retrieval, grounding, and answer synthesis. GEO improves the conditions it can influence at each step and helps diagnose failures specific to that step.
  • Citation vs Mention vs Link
    Concept AI answers can credit a source through a citation, a mention, or a link. These outcomes are not interchangeable: using a source and crediting it are separate events, and each form is measured differently, influenced by different factors, and valuable for different reasons.
  • Generative Engine
    Concept A generative engine retrieves sources, uses them to ground an LLM, and synthesizes a written answer that may credit those sources. Unlike a search engine, it composes an answer instead of returning documents. This is the system that GEO aims to influence.
  • Generative Engine Optimization
    Concept Generative Engine Optimization (GEO) improves the chances that AI systems such as ChatGPT, Perplexity, Google AI Overviews, and Gemini will retrieve your content, use it to support an answer, and cite or mention it. GEO extends SEO; it does not replace it.
  • GEO ROI Models
    Concept Generative search separates value from the click, so traditional ROI models miss much of GEO's impact. A three-currency framework measures citation value, substituted traffic value, and brand authority value across B2B SaaS, B2C e-commerce, and media.
  • LLMO vs GEO
    Concept LLMO and GEO are mostly the same thing. They share a goal and nearly identical definitions in practice. LLMO emphasizes the model rather than the answer; only training-corpus inclusion makes it distinct, and GEO excludes that largely uncontrollable work.
  • SEO vs GEO
    Concept SEO and GEO share the same foundation: crawlability, real expertise, clear structure, and authoritative mentions. SEO aims for a ranked link that earns a click; GEO aims for a citation or mention in a synthesized answer. GEO builds on SEO rather than replacing it.
  • Zero-click Search
    Concept A zero-click search ends on the results page without a visit to a source. Most Google searches already ended without a click before generative answers further increased zero-click behavior. For GEO, value shifts from earning a click to earning a citation or mention.

Signals

  • AI Content Detection
    Concept AI engines down-weight patterns associated with low-effort or scaled production, whether AI tools were involved or not. Classifier scores do not control citation or visibility and should not be treated as GEO signals.
  • Brand Mentions
    Concept An unlinked brand mention strengthens the model's prior about an entity. Repeated references across the web can influence future answers without providing a link, so mentions and links serve different purposes.
  • Citability
    Concept Citability is the structural property that determines whether an AI system can lift a retrieved passage intact into its answer. It is the structural factor in grounding; E-E-A-T is the trust and authority factor.
  • Content Freshness
    Concept Content freshness combines a page's age (recency) with whether its claims remain true (currency). AI engines favor recent content, but the average AI-cited page is about 2.9 years old. Freshness depends on the query; it is not a reason to publish every week.
  • E-E-A-T
    Concept E-E-A-T stands for Experience, Expertise, Authoritativeness, and Trustworthiness. Google uses it as a framework for quality raters, not an algorithmic score. For GEO, it addresses whether the source behind a retrieved, liftable passage is trustworthy enough to use.
  • Entity Recognition
    Concept Entity recognition is how an AI engine maps a brand, product, or author name to the correct known entity. Accurate resolution lets mentions, citations, and markup reinforce the intended entity; ambiguity can send that credit elsewhere or cause it to be discarded.
  • GEO Spam & Manipulation
    Concept Adversarial GEO now appears in both published research and commercial practice. Every published attack succeeds in a single-attacker study, but competition consistently shrinks or reverses the gains. Two of the four documented attack surfaces lie outside your website.
  • Knowledge Graph Presence
    Concept Knowledge graph presence means having an entity node in a graph an AI engine trusts, such as Wikipedia, Wikidata, or the Google Knowledge Graph. It strengthens the model's prior and aids resolution, but it does not cause citations or become credible through self-declaration.
  • Multilingual GEO
    Concept Multilingual GEO examines how AI retrieval, entity matching, passage extraction, and trust signals change across languages. The answer loop stays the same, but source pools, entity binding, chunk structure, and trust pools vary. Source-pool differences matter most.
  • Multimodal Signals
    Concept Multimodal signals help AI engines interpret non-text assets such as images, video, audio, and charts. In 2026, text associated with an asset, including alt text, captions, transcripts, and schema, still matters more than pixel-level visual analysis in both retrieval modes.

Infrastructure

  • AI Crawlers
    Concept AI crawlers fall into three categories with different access consequences: training, retrieval, and user-triggered agents. Access decisions should be made by category, because blocking retrieval crawlers to prevent training can remove a site from AI answers.
  • ChatGPT-User
    Concept ChatGPT-User fetches a page when it is needed during a live ChatGPT interaction. OpenAI says robots.txt may not apply and confirms that the token does not determine ChatGPT Search inclusion. Blocking it can prevent the page from appearing in a reader's answer.
  • Core Web Vitals (LCP/INP/CLS)
    Concept Core Web Vitals (LCP, INP, and CLS) are Google ranking signals, not AI-engine signals. Their GEO effect is direct for Google AI Overviews, partial for Bing Copilot, and negligible for ChatGPT Search, Perplexity, and Claude. AI crawler performance is a separate issue.
  • Google-Extended
    Concept Google-Extended is a robots.txt control token, not a crawler. It governs specified Gemini training and grounding uses without affecting Search inclusion or ranking. Search AI appearance and grounding are governed by a separate Search Console control that is still rolling out.
  • GPTBot
    Concept GPTBot is OpenAI's training crawler. It collects content that may be used to train future models, but it does not support ChatGPT Search or live answers. Blocking it does not reduce citations, but doing so may mean forgoing a future benefit that cannot be measured or attributed.
  • JSON-LD
    Standard JSON-LD is one of three ways to express Schema.org markup. Google recommends it because it stays separate from visible HTML and is easy to maintain. Index-integrated AI parses the markup as structured data, while live-fetch chatbots read it as page text.
  • llms.txt
    Standard llms.txt is a convention proposed by Answer.AI in 2024: a curated markdown file at /llms.txt that identifies the pages an LLM should read first. Sites are publishing it, but no major AI vendor confirms using it. It is a low-cost precaution, not a citation channel.
  • robots.txt
    Standard RFC 9309 defines robots.txt as a voluntary request, not access control. Compliant AI crawlers follow it as documented; noncompliant and spoofed agents may not. Set rules separately for training, retrieval, and user-triggered agents, then verify access at the network layer.
  • Schema.org for AI
    Concept Schema.org markup is not a ranking or citation signal. It helps AI systems distinguish entities and parse pages. Index-integrated AI uses it, while live-fetch chatbots read JSON-LD as plain text. Markup can make an entity resolvable, but not a passage citable.
  • Sitemap & IndexNow
    Standard Sitemap.xml is a pull protocol from 2005. Microsoft and Yandex introduced the push-based IndexNow protocol in 2021. Both affect AI visibility through host indexes: Google AI Overviews via Google and Bing Copilot via Bing. ChatGPT, Perplexity, and Claude use neither directly.

Practice

  • GEO Metrics
    Concept These 10 core GEO KPIs cover definitions, formulas, SEO equivalents, and the different ways Profound, Otterly, Ahrefs, BrightEdge, and Similarweb define them. The set is a GEO Wiki synthesis, not an industry standard.