GEO Wiki: Terms of AI Visibility Explained
From aeo® to citability: The ainomics® GEO Wiki defines all the central terms around Generative Engine Optimization and AI visibility — compact and citable.
All terms (A–Z):
- 5 stages of AI visibility
- 301 redirect
- aeo® — ainomics engine optimization
- AEO — Answer Engine Optimization as a generic term
- aeo® loop
- aeo® audit — Audit & Diagnosis
- aeo® KPIs
- aeo® model
- Freshness vs. authority
- Alt text
- Anchor text
- Answer block
- Author byline
- Backlink
- Baseline
- Preferred content formats
- Branded prompts
- Breadcrumb / BreadcrumbList
- Canonical tag
- ChatGPT / ChatGPT Search
- Chunk
- Citation rate
- Claude
- Core Web Vitals
- Crawling and indexing
- Customer journey prompts
- DefinedTerm markup
- Definition-first principle
- Domain authority
- Three truths of aeo®
- Duplicate content
- E-E-A-T
- Earned content
- Entity
- Mention
- FAQPage markup
- Featured snippet
- Funnel mapping
- Generative engine / answer engine
- Generative Engine Optimization — GEO
- GEO vs. SEO copywriting
- GEO landing page
- Google AI Overviews
- Google Gemini
- Google Search Console / Bing Webmaster Tools
- Grounding
- Hallucination
- hreflang
- JSON-LD
- Keyword
- AI crawlers — GPTBot, ClaudeBot, PerplexityBot and co.
- Knowledge cutoff
- Knowledge graph
- Consistent terminology
- Context window
- Large Language Model — LLM
- LLMO — Large Language Model Optimization
- llms.txt
- Local SEO
- Long tail
- Microsoft Copilot
- Monitoring layers
- Narrative architecture
- Non-branded prompts
- On-page and off-page optimization
- Owned content
- PageRank
- Perplexity
- Persona mapping
- Person authority
- Prompt
- Prompt set
- Source authority
- Ranking
- Retrieval / RAG
- robots.txt
- SEO — Search Engine Optimization
- SERP — search engine results page
- Server-side rendering — SSR
- Share of model
- Social proof stacking
- Structured data / schema markup
- Search intent
- Title tag and meta description
- Token
- Training vs. retrieval
- Whitepaper and case study
- Wikipedia and Reddit as GEO sources
- Wikipedia stewardship
- XML sitemap
- YMYL — Your Money or Your Life
- Citability
- Citation
5 stages of AI visibility
AI visibility grows stage by stage: from first mention through correct description and active recommendation to linking and finally to being a regularly cited source. Each stage is measured by a clearly defined KPI.
Related terms:
301 redirect
Permanently forwards a URL to a new one and transfers its authority. Mandatory in every rebuild: references stored by AI systems must not lead nowhere.
aeo® — ainomics engine optimization
aeo® is the method developed by ainomics GmbH in Vienna to make companies visible in generative AI systems and to actively steer how they are represented. It combines Generative Engine Optimization with digital consulting and communications in a steering loop — from persona mapping and prompt sets to KPIs and continuous monitoring. GEO optimizes for being found; aeo® optimizes for being understood.
Related terms:
AEO — Answer Engine Optimization as a generic term
Another industry term for optimizing for answer engines. Not to be confused with aeo® by ainomics®: aeo® is a proprietary method that integrates consulting, communications and a steering loop beyond pure optimization.
Related terms:
aeo® loop
The continuous fourth phase of the method: monitoring across several observation layers, KPI reporting, Wikipedia stewardship and ongoing optimization of content and structured data.
Related terms:
- aeo® — ainomics engine optimization
- aeo® audit — Audit & Diagnosis
- Monitoring layers
- Wikipedia stewardship
aeo® audit — Audit & Diagnosis
The first phase of every aeo® project: it measures whether and how a company appears in AI answers, based on persona mapping and a dedicated prompt set, measured per AI system. The deliverable is a detailed action plan. Duration approx. 6 weeks, 6 consulting days one-off.
Related terms:
aeo® KPIs
Each visibility stage gets its own KPI — mention rate, accuracy of description, recommendation rate or citation rate — measured per AI system and prompt group against baseline and competitors.
Related terms:
aeo® model
The aeo® model unites three disciplines: GEO for technical and content findability, digital consulting for strategy and steering, and communications & PR for trust and third-party sources. Only their interplay makes AI visibility plannable, measurable and commercially effective.
Related terms:
- aeo® — ainomics engine optimization
- Generative Engine Optimization — GEO
- Owned content
- Earned content
Freshness vs. authority
For news, laws, prices and market trends AI systems retrieve current web sources — freshness decides. For timeless knowledge, definitions and methods, the authority anchored in the model counts. A GEO strategy must serve both.
Related terms:
Alt text
Describes an image in text form — for accessibility and machines. GEO rule: diagrams need alt texts that explain the core message in one sentence, otherwise their content is lost to AI systems.
Anchor text
The clickable wording of a link. Descriptive anchors such as to the aeo® audit instead of phrases such as learn more help users, search engines and LLMs classify the link target.
Answer block
A self-contained section of 50–300 words under a clear heading that answers a question directly — the answer in the first sentence. Answer blocks are the basic unit of GEO-ready content, matching the chunk processing of AI systems.
Related terms:
Author byline
Names the author of a piece with role and profile link. It connects content with person authority — a building block of E-E-A-T and a distinct signal for source credibility.
Related terms:
Backlink
A link from a third-party website to your own — the classic SEO authority signal. For GEO, mentions without links also count: repeated brand mentions in the right topical context strengthen the associations LLMs learn.
Related terms:
Baseline
The documented initial measurement of AI visibility before any measures — per prompt, prompt group and AI system. Since there are no accepted industry norms for citation rates, your own baseline is the only reliable reference point.
Related terms:
Preferred content formats
Four formats are cited most: definitions for term queries, studies and data as verifiable facts, how-to content as actionable guidance, and FAQs with real questions and concise answers.
Related terms:
Branded prompts
Questions that name the company or brand directly — about services, reviews or comparisons. They identify what makes the company unique and test whether AI systems describe the brand correctly.
Related terms:
Breadcrumb / BreadcrumbList
Breadcrumbs show a page’s position in the site hierarchy and become machine-readable via BreadcrumbList markup. They strengthen semantic structure — AI systems understand how a subpage fits into the topic architecture.
Related terms:
Canonical tag
Names the authoritative URL of a piece of content and prevents duplicates from splitting authority. Every page should be self-referencing canonical.
Related terms:
ChatGPT / ChatGPT Search
ChatGPT by OpenAI works model-based or with web access and is strong at explanatory, summarizing and advisory answers. ChatGPT Search adds live web sources with citations — usage has been growing rapidly across Europe for years.
Related terms:
Chunk
The text segment an AI system processes and cites as a unit — typically about 100–500 tokens. Hence: answer blocks of 50–300 words under a clear heading, understandable on their own, are adopted in full most often.
Related terms:
Citation rate
The share of relevant AI answers in which a source is named or cited. There are no generally accepted industry norms — your own baselines and competitor comparison over time are what counts.
Related terms:
Claude
Anthropic’s LLM assistant, focused on long contexts and careful answers. Relevant for GEO as a standalone answer system with its own crawler ClaudeBot — and as proof that every engine has its own data sources and selection logic.
Related terms:
Core Web Vitals
Google’s metrics for loading, interactivity and visual stability. Primarily an SEO signal, but indirectly relevant for GEO: fast, stable pages are crawled more reliably and used more often as sources.
Related terms:
Crawling and indexing
Crawling is the automated reading of a website by bots, indexing the inclusion of its content in the searchable corpus. Both precede any visibility — what is not crawled and indexed can neither rank nor be cited.
Related terms:
Customer journey prompts
Questions along the decision journey — from first orientation through comparison to purchase. They connect the prompt set with funnel mapping and reveal at which stage visibility is missing.
Related terms:
DefinedTerm markup
The schema.org type for term definitions, bundled in a DefinedTermSet. Every wiki entry gets this markup — making the glossary readable as a coherent term system and ainomics® recognizable as the defining source.
Related terms:
Definition-first principle
Every section starts with the answer or definition, followed by context, evidence and example — not the other way round. Definitions are the most-cited content format.
Related terms:
Domain authority
The cross-topic strength of a domain as estimated by SEO tools. AI systems assess authority context-specifically per topic — a strong domain helps but does not replace repeated association with the specific topic field.
Related terms:
Three truths of aeo®
First: search becomes a question — people trust AI answers directly. Second: narrative is not visibility — a company can be known to LLMs and still be described incorrectly. Third: communication without control — without monitoring, the AI alone decides what image of a company emerges.
Related terms:
Duplicate content
Identical content under several URLs. It splits authority and confuses source selection — canonical tags and a clear page architecture with one main question per page are the remedies.
Related terms:
E-E-A-T
Experience, Expertise, Authoritativeness, Trustworthiness — the quality frame for trustworthy sources. For AI visibility this means: recognizable authors, proven expertise, consistent company data and independent third-party confirmation.
Related terms:
Earned content
Third-party content about the brand: press coverage, expert articles, study citations, podcasts, reviews. As independent confirmation it strongly builds source authority — the indirect, slower but more sustainable lever.
Related terms:
Entity
A uniquely identifiable object — a company, person, product or place. AI systems prefer entities over mere keywords: consistent names, structured data and links — for example Wikidata or LinkedIn — make a brand readable as an entity.
Related terms:
Mention
The naming of a brand in third-party content — with or without a link. In AI search, mentions gain weight over backlinks because language models learn from text: frequent, consistent mentions in topical context shape the brand’s image.
Related terms:
FAQPage markup
The schema.org type for question-and-answer sections. It makes FAQs machine-readable — ideal for GEO, because real user questions with concise answers are exactly what generative systems look for and adopt.
Related terms:
Featured snippet
The highlighted direct answer at the top of Google search — the precursor of the AI answer. Content that wins snippets — precise answer blocks under a clear question — is usually GEO-ready too.
Related terms:
Funnel mapping
Funnel mapping assigns prompts to decision stages — from newcomer to decision-maker. Every funnel stage needs its own prompts and its own GEO KPI, because AI systems select different sources per question type.
Related terms:
Generative engine / answer engine
An AI system that produces directly formulated answers to questions instead of link lists — based on trained knowledge and, depending on the system, live-retrieved web sources. Examples: ChatGPT, Perplexity, Google AI Overviews, Gemini, Copilot.
Related terms:
Generative Engine Optimization — GEO
GEO covers all measures that make content visible and citable in generative AI systems such as ChatGPT, Perplexity, Gemini or Copilot. Established in research since 2024, it is the counterpart to SEO in AI search: GEO asks why a system chooses a source, not how a page ranks.
Related terms:
- aeo® — ainomics engine optimization
- SEO — Search Engine Optimization
- Generative engine / answer engine
- GEO landing page
GEO vs. SEO copywriting
Classic SEO copywriting focuses on keywords, volumes and rankings. GEO copywriting explains relationships — term, context and function directly in the text, with consistent terminology and entities instead of keyword repetition. The goal is adoption into the answer, not placement.
Related terms:
GEO landing page
Answers one unambiguous main question that addresses the user directly. No universal template — shared principles: a clear main question, a semantic structure with H1 for the topic and H2 for sub-questions, citable answer blocks and topic-specific FAQs.
Related terms:
Google AI Overviews
The AI-generated answer summaries above classic Google results. They draw on web sources and fundamentally change click behaviour: the answer comes before the links — whoever is not cited there loses visibility.
Related terms:
Google Gemini
Google’s AI model — as a standalone assistant and as the engine behind AI Overviews. Its closeness to the Google index means classic crawling and structure signals feed strongly into AI answers.
Related terms:
Google Search Console / Bing Webmaster Tools
The search engines’ free diagnostic tools: indexing status, crawl errors, queries, structured data. Indispensable for GEO because Bing underpins Copilot and Google underpins AI Overviews and Gemini.
Related terms:
Grounding
Anchoring an AI answer in concrete, verifiable sources — the antidote to hallucination. Systems with strong grounding — such as Perplexity — show their sources visibly; serving as a grounding source wins visibility and trust.
Related terms:
Hallucination
A factually wrong but plausibly worded AI answer. For brands, hallucinations are a reputation risk — aeo® counters them with correct, consistent, machine-readable sources and continuous monitoring of brand representation.
Related terms:
hreflang
Links the language versions of a page — for example German and English — so users and systems get the right version. Precondition: both versions exist as standalone, mirrored URLs.
Related terms:
JSON-LD
The recommended format for structured data: a script block in the page head describing content according to schema.org without changing the visible HTML. The markup must match the visible content.
Related terms:
Keyword
The search term a user types into a search engine. In AI search, keyword density loses importance: AI systems understand contexts and topics, not just terms — entities and context beat repetition.
Related terms:
AI crawlers — GPTBot, ClaudeBot, PerplexityBot and co.
The bots with which AI providers capture web content — separated into training bots such as GPTBot and Google-Extended, and search and retrieval bots such as OAI-SearchBot and PerplexityBot. robots.txt lets you decide per bot; whoever wants visibility opens at least the search bots.
Related terms:
Knowledge cutoff
The point in time up to which a language model was trained — everything after it the model only knows via retrieval. Appearing once in training data is not enough: current, crawlable content keeps a brand’s representation up to date.
Related terms:
Knowledge graph
A structured network of entities and their relationships — such as the Google Knowledge Graph or Wikidata. Being correctly recorded there makes search and AI systems recognize, classify and represent a brand more reliably.
Related terms:
Consistent terminology
The same terms for the same things — across website, press, social media and third-party sources. Inconsistent naming dilutes the entity; consistency strengthens recognition, trust and the associations LLMs learn.
Related terms:
Context window
The amount of text an LLM can process at once — question, conversation and retrieved sources combined. Compact, self-contained content fits the window better and is more likely to be adopted in full.
Related terms:
Large Language Model — LLM
An AI model trained on vast amounts of text that understands and generates language — the technological basis of ChatGPT, Claude, Gemini and Copilot. LLMs answer from trained knowledge and, depending on the system, from live-retrieved web sources.
Related terms:
LLMO — Large Language Model Optimization
A neighbouring term to GEO: optimizing content and signals for how large language models learn, store and reproduce a brand. Terminology in this young field is not yet settled; GEO is the academically established term.
Related terms:
llms.txt
A markdown file in the website root giving AI systems a curated overview of a site’s most important content — a table of contents for LLMs, analogous to robots.txt for crawlers.
Related terms:
Local SEO
Optimizing visibility for location-based searches — via directories, Google Business Profile and consistent name, address and phone data. AI assistants increasingly recommend local providers — consistent directory data is the foundation — local GEO.
Related terms:
Long tail
Long, specific queries with low volume but high intent. Prompts are born long-tail queries — whoever answers specific questions precisely wins disproportionately in AI search.
Related terms:
Microsoft Copilot
Microsoft’s AI assistant in Windows, Office and Bing. It combines LLM answers with the Bing index; visibility in Bing — maintained via Bing Webmaster Tools and indexable content — is the basis for visibility in Copilot.
Related terms:
Monitoring layers
No single tool captures GEO completely. aeo® combines several observation layers: the technical layer showing why a source is selected and whether the brand is cited, the brand and prompt perspective showing how the brand is described, and competitive benchmarks over time.
Related terms:
Narrative architecture
The deliberate construction of a company’s core messages across all channels: which statements AI systems should learn, in which wording, supported by which evidence. Consistent terminology is the precondition for LLMs storing a coherent picture.
Related terms:
Non-branded prompts
Generic questions without brand names — such as asking for the best provider or the right method. They measure whether a brand appears in generic AI answers at all: the hardest and most valuable form of AI visibility.
Related terms:
On-page and off-page optimization
On-page covers all measures on your own site — content, structure and technology. Off-page covers everything outside — links, mentions and press. The GEO equivalent: owned content as the direct lever, earned content as the indirect one.
Related terms:
Owned content
All content on your own channels: website, blog, whitepapers, case studies, press section, social profiles. The direct lever with full control — quick to deploy, moderate as an LLM signal; what counts are narrative architecture, structured data and GEO-ready formats.
Related terms:
PageRank
Google’s historic algorithm computing authority from the web’s link structure. No publicly known mechanism of this kind exists for LLMs — AI systems use many signals to assess relevance and authority. GEO asks about source selection, not placement.
Related terms:
Perplexity
Built as an answer engine with source logic: sources sit visibly at the centre of every answer. That makes Perplexity especially revealing for GEO analysis — it shows directly which sources a system selects for which question.
Related terms:
Persona mapping
Persona mapping describes who asks an AI system questions: in which role, with which prior knowledge, context and goal. Since LLM users ask rather than search, it is the foundation of every robust prompt set.
Related terms:
Person authority
The perceived topical competence of an author. Consistent topic coverage, regular publications and visible profiles on LinkedIn and in author bylines strengthen the company’s credibility — a distinct LLM signal.
Related terms:
Prompt
The question or instruction a user gives an AI system — in natural language, often with a role, context and goal. Prompts are the GEO equivalent of search terms, but longer, more specific and conversational.
Related terms:
Prompt set
A structured collection of real user questions used to measure a company’s AI visibility, covering branded, non-branded and customer journey prompts. It is the measurement basis of every aeo® project.
Related terms:
Source authority
From an AI system’s perspective, authority arises from the repeated association of a source with a topic — via your own website and independent third parties. It is context-specific and does not transfer automatically between topics.
Related terms:
Ranking
A page’s position in organic results. In AI answers there is no ranking in the classic sense — there is selection: a source is cited or it is not. GEO KPIs therefore measure mention, description and citation instead of positions.
Related terms:
Retrieval / RAG
Retrieval-augmented generation: an AI system’s live access to current web sources while answering. For news, prices or laws retrieval decides — for timeless knowledge the authority anchored in the model counts. GEO must serve both paths.
Related terms:
robots.txt
Controls which crawlers may read a website — including AI bots. Opening increases visibility and findability in AI answers; closing protects content but reduces the chance of being cited. Decide deliberately per bot.
Related terms:
SEO — Search Engine Optimization
All measures to rank well in classic search engines: technology, content and links. Relative to GEO: SEO optimizes for rankings on results pages, GEO for selection as a source in AI answers — they share the technical foundation.
Related terms:
SERP — search engine results page
The results page with organic hits, ads and increasingly AI elements such as AI Overviews. The transformation of the SERP — answer before links — is the most visible expression of the shift from SEO to GEO.
Related terms:
Server-side rendering — SSR
Server-side rendering means the server delivers finished HTML instead of building content in the browser via JavaScript. Crucial for GEO: many AI crawlers do not execute JavaScript — content that appears only after a click or scroll does not exist for them.
Related terms:
Share of model
A brand’s share of an AI system’s answers in a topic field — analogous to classic share of voice. Measured via the prompt set: in how many answers does the brand appear compared with competitors?
Related terms:
Social proof stacking
Making one statement, brand or person visible across several channels at once — LinkedIn, interviews, podcasts, expert articles, press, talks. Several independent sources make topical classification easier for AI systems and build trust.
Related terms:
Structured data / schema markup
Structured data based on schema.org, usually embedded as JSON-LD, makes content machine-readable: companies, people, services, FAQs, events, reviews. Schema markup makes relationships between entities explicit — a central signal for how AI systems capture information.
Related terms:
Search intent
What a user wants to achieve with a search: inform, compare, buy, navigate. GEO extends the concept to the prompt: questions already carry role, context and goal — content must answer the underlying intent.
Related terms:
Title tag and meta description
The title tag is the page title in results — main term, benefit and brand in roughly 60 characters. The meta description is the text below — around 150 to 160 characters, phrased as an answer. Both shape click-through and comprehension — AI systems read them as the page summary.
Related terms:
Token
The smallest processing unit of a language model — roughly a syllable to a short word. Relevant for GEO because AI systems process content in chunks of about 100–500 tokens: answer blocks must be self-contained at that scale.
Related terms:
Training vs. retrieval
AI answers draw on two sources: the model’s trained knowledge, which is frozen at the knowledge cutoff, and live-retrieved web content. Long-term authority must seep into training — via Wikipedia, established media and consistent presence — while current content must be crawlable and citable.
Related terms:
Whitepaper and case study
Owned-content formats with high citability: they deliver data, methods and verifiable results — exactly the evidence AI systems preferentially adopt.
Related terms:
Wikipedia and Reddit as GEO sources
Both influence AI answers disproportionately: Wikipedia supplies structured information and entities and strengthens source trust; Reddit bundles authentic user questions and experience whose community consensus reinforces relevance.
Related terms:
Wikipedia stewardship
The ongoing, rules-compliant care of a company’s Wikipedia presence: auditing existing articles, sourcing, watching talk pages. Wikipedia provides AI systems with structured entities and trust — and influences AI answers disproportionately.
Related terms:
XML sitemap
Lists all indexable URLs of a website and is submitted to Google Search Console and Bing Webmaster Tools. It speeds up discovery of new and changed pages — the basis for retrieval systems knowing current content.
Related terms:
YMYL — Your Money or Your Life
Topics with direct impact on health, finances or safety. In YMYL industries — healthcare, banking, insurance — search and AI systems apply particularly strict source standards; correct representation is business-critical.
Related terms:
Citability
How well content can be adopted by AI systems as a source: clear statements instead of phrases, concrete facts, freshness, traceable sources and consistent terminology — structured in self-contained answer blocks.
Related terms:
Citation
When an AI system names or links a source in its answer. Citations are the hardest currency of AI visibility: they carry authority, generate traffic and make the source more likely to be chosen again.
Related terms:
