Reference
GEO, AEO and LLM SEO Glossary - no-nonsense terms
This sector invents vocabulary faster than it can define it accurately. Here's what each term really means, without marketing inflation - so you use the right word and know exactly what it implies.
Fundamental concepts
GEO - Generative Engine Optimization
Set of practices that make a brand, product or website more likely to be cited by generative language models - such as ChatGPT, Gemini, Perplexity or Claude - when they answer user questions. Unlike SEO, there is no public ranking to consult - you only know if a brand appears by systematically triggering prompts and analyzing the responses.
AEO - Answer Engine Optimization
Practice optimizing to appear in direct answers from traditional search engines - featured snippets, answer panels, and “people also ask” blocks. There is significant technical overlap with GEO (structure, clarity, FAQ), but the evaluation mechanisms and target engines are different: AEO serves classical search engines, GEO serves generative LLMs.
LLM Search
Search layer where a language model does real-time web searching before responding - for example, ChatGPT with active search or Perplexity. It differs from the organic layer (which the model already knows from its training) because it depends on the traditional SEO of the website consulted at the time of the search.
RAG - Retrieval-Augmented Generation
Technology that allows a language model to access external and updated information in real time, instead of relying solely on knowledge fixed in its training. It is the technical engine behind features such as web search in AI assistants and business systems that query their own knowledge bases.
Metrics and scores
GEO Score
Index from 0 to 100 that measures a brand's organic visibility in LLM responses, combining frequency of mention, relative position in the response and citation quality. Calculated from organic responses only - no web search or injected RAG context - to keep the number stable and comparable. Full methodology here.
RAG-Readiness
Measure of how well a website's content is structured to be retrieved effectively by RAG systems - the technology that allows LLMs to access up-to-date information. A website can have high RAG-Readiness but low GEO Score: this indicates that the content exists and is well prepared, but has not yet been incorporated into the organic knowledge of the models.
IPD-IA - AI Digital Readiness Index
Independent framework that analyzes the level of digital preparation of Portuguese company websites for AI response engines, evaluating five pillars: technical structure, semantic structure, AI preparation, external authority and positioning strategy. See results by sector.
Technical signals
llms.txt
File published in the root of a domain that guides AI crawlers about the site's structure and priority content - the equivalent, for language models, of what the sitemap.xml it is for traditional search engines.
ai.json
Structured metadata file at the root of a domain, designed for direct consumption by AI systems - describes the brand, services and context in a machine-readable way, complementing the structured data already present in HTML.
Schema Markup / Structured Data
Standardized vocabulary (Schema.org) inserted into the code of a page to explicitly describe entities, products, services or events, so that search engines and AI systems can interpret the content without ambiguity.
Entity Clarity
Degree to which an LLM can unambiguously identify what a brand is, does, and for whom - without relying on inference or contradictory data across platforms. It is one of the most determining factors in the likelihood of a brand being cited correctly.