Glossary: every GEO and generative AI term

GEO (Generative Engine Optimization)

The set of practices aimed at optimizing content and digital presence to be cited as a source by generative AI engines (ChatGPT, Claude, Gemini, Perplexity) when they build an answer.

AEO (Answer Engine Optimization)

Optimization aimed at becoming THE direct answer to a specific question, whether in an AI chatbot or in a traditional search engine's enriched results (AI Overviews, voice assistants).

AI visibility

The fact, for a brand, product or piece of content, of being mentioned in the answers generated by a generative AI — the equivalent of "ranking" in SEO, but at the scale of a single answer rather than a results page.

AI answer / generative answer

The text produced by an LLM in response to a question, built from the data it was trained on and, in some cases, real-time search.

LLM (Large Language Model)

A large language model, like the ones powering ChatGPT (OpenAI), Claude (Anthropic), Gemini (Google) or Perplexity.

Visibility score

A summary indicator that reflects a brand's, category's or keyword's presence across all tested prompts and engines.

Share of voice

The proportion of mentions a brand gets relative to its competitors, across the same set of tested questions.

Prompt

A question or query sent to an LLM to test whether it mentions a given brand, category, keyword or product.

Persona

A fictional user profile (age, location, language, interests, buying behavior) used to simulate different types of questions and detect visibility gaps across audiences.

Semantic scoring

The evaluation of how "AI-friendly" a product page or piece of content is, based on criteria like structure, structured data, the presence of an FAQ, or out-of-context readability.

Structured data (schema.org / JSON-LD)

Technical markup added to a web page to explicitly describe its content (product, price, reviews, FAQ...), making it easier for search engines and AIs to understand.

Detailed analysis

An in-depth report generated from visibility results: a summary, dominant and absent players, competitive comparison, risks and opportunities, prioritized recommendations.

Credits

The unit of consumption for analyses on Vurto; each analysis consumes a variable number of credits depending on its type and how many engines are queried.

RAG (Retrieval-Augmented Generation)

Retrieval-augmented generation is a technique that lets an LLM fetch information from a database or the web at the moment it answers, rather than relying solely on its training data. This is the mechanism that lets tools like Perplexity, or ChatGPT's search features, cite recent sources.

Fan-out query

A technique by which a generative engine breaks a complex question into several sub-queries sent in parallel, then synthesizes the results into a single answer. A single question from a user can therefore trigger several distinct searches behind the scenes.

Reranking / RRF (Reciprocal Rank Fusion)

Reranking refers to reordering a set of already-retrieved results based on their actual relevance to the question asked. RRF (Reciprocal Rank Fusion) is a common method for merging several result rankings from different sources or search methods into a single final ranking.

Chunking

Splitting a long document into smaller fragments ("chunks") processed and indexed independently. This technique, central to RAG systems, explains why a well-structured page with self-contained sections is more likely to be used correctly than a long text without clear breaks.

E-E-A-T

Stands for Experience, Expertise, Authoritativeness, Trustworthiness — a set of criteria Google uses to assess content quality, notably through its quality rater guidelines. While not a direct, measurable ranking factor, these criteria influence how trustworthy content is judged to be, by both search engines and generative AIs.

Essential HTML tags (Title, Meta description, Hn, Canonical, Hreflang, Open Graph)

A set of technical elements placed in a page's code to describe its content to engines and social networks: the Title (displayed in results), the Meta description (summary shown under the title), Hn tags (heading hierarchy, H1 to H6), the Canonical tag (indicating the reference version of a duplicated page), the Hreflang tag (signaling the language versions of the same page), and Open Graph tags (controlling how a page's preview looks when shared on social media).

Core Web Vitals

A set of metrics defined by Google to measure a page's user experience: how fast the main content loads, how responsive it is to interactions, and how visually stable it is while loading. These metrics are used as a ranking factor in SEO and indirectly contribute to how well a site can be crawled.

robots.txt

A file placed at the root of a site that tells bots (search engines, AI crawlers) which parts of the site they are and aren't allowed to browse. It doesn't guarantee content will never be used — a bad-faith bot can ignore it — but it remains the standard way to communicate crawling preferences.

llms.txt

A file proposed as an emerging standard, placed at the root of a site, meant to give language models a structured, condensed overview of the site's content — somewhat like a sitemap designed specifically for AIs rather than traditional search engines. Its adoption remains uneven across engines as of today.

XML Sitemap

A file listing a site's pages intended for indexing, along with additional information such as their last modification date. It makes pages easier for bots to discover, especially on large sites or ones with a complex internal linking structure.

AI crawlers (GPTBot, ClaudeBot, PerplexityBot...)

The bots used by generative AI companies to crawl the web, either to train their models or to power their real-time search features. GPTBot (OpenAI), ClaudeBot (Anthropic), and PerplexityBot (Perplexity) are examples; their activity can be tracked in a site's server logs, and allowed or blocked via the robots.txt file.

AI Overviews

AI-generated summaries that appear at the top of certain Google search results, ahead of standard organic results. They represent a form of AEO built directly into traditional search, and can reduce traffic to cited sites if the user finds their answer without clicking through.

Hub & Spoke architecture

A site organization model in which a pillar page ("hub") centralizes an overview of a topic, linked to several more specific pages ("spokes") that each dig into a sub-topic. This structure helps engines understand the hierarchy and relationships between content on the same subject.

Pillar page / spoke page

The pillar page is the reference content on a broad topic, designed to cover the essentials and link out to more detailed content. A spoke page covers a specific sub-topic in depth and links back to its pillar page for broader context — together, they form the hub & spoke architecture.

Entity (SEO)

A person, place, organization, product, or concept identified as a unique thing by search engines, independent of the exact words used to describe it. Working on entity recognition means helping engines clearly associate a brand with what it represents, beyond just the text on its pages.

Search intent

The actual motivation behind a query: looking for information, comparing options, making a purchase, or finding a specific site. Identifying the intent behind a question is a prerequisite for creating relevant content, whether for SEO or to be picked up in a generative answer.

Link building

The set of practices aimed at earning external links pointing to your site from other domains. Historically central to SEO as an authority signal, it retains a role in GEO insofar as it contributes to a brand's overall reputation and the mentions that stem from it.

Internal linking

The set of links connecting a site's pages to one another. Consistent internal linking makes crawling easier, distributes authority across pages, and helps engines understand how a site's content relates thematically.

Local SEO

The set of practices aimed at improving a business's visibility for geolocated searches, relying notably on the Google Business Profile listing, consistent business information (NAP), and customer reviews.

Multimodal content

Content that isn't plain text — images, video, audio, infographics. For it to be usable by engines and generative AIs, it needs to come with descriptive text (alt text, captions, transcripts).

MCP (Model Context Protocol)

A standardized protocol that lets an AI assistant connect to external tools and data sources in a structured way. At Vurto, it's offered on the Business and Agency plans, letting third-party systems query visibility data directly.

To see these concepts in practice, head to prompts & visibility tracking and dashboards & scoring.