AI Visibility Glossary

The measure of how often and prominently a brand appears in responses generated by AI models like ChatGPT, Claude, Gemini, and Perplexity. AI visibility is the AI-era equivalent of search engine rankings.

AI-generated summaries that appear at the top of Google Search results pages, powered by Google Gemini. AI Overviews synthesize information from multiple web pages and display a direct answer before traditional blue links.

The practice of optimizing content and online presence to appear in AI-generated responses. GEO is the evolution of SEO focused on being mentioned or recommended by large language models (LLMs) rather than ranked in traditional search results.

A content strategy focused on making your brand the recommended answer to questions asked to AI assistants and answer engines like Perplexity. AEO overlaps with GEO but focuses specifically on question-answer formats.

A group of related prompts and queries organized around a central theme or product category. In Mentova, a topic is the set of prompts used to measure brand visibility within a specific market segment or use case.

A simulated user profile injected into AI prompts to test how different audience segments influence brand recommendations. In Mentova, audience personas (e.g., 'Alex, 25-year-old product manager') are used to understand how AI visibility varies across user types.

The percentage of AI-generated brand mentions that belong to your brand compared to competitors. If 5 brands are mentioned across 100 AI responses in a category and your brand appears 30 times, your AI SoV is 30%.

The percentage of AI prompts in which a specific brand is mentioned at least once. A mention rate of 40% means the brand was cited in 40 out of 100 prompts sent to an AI model on a given topic.

A metric (0–100) measuring how much a specific web source shapes the outputs of AI models. Sources with high influence scores (like Wikipedia, Reddit, G2) are frequently cited in AI training data and responses. Building presence on high-influence sources is a key GEO strategy.

An occurrence of a brand name, product name, or company identifier within an AI-generated response. Brand mentions are the core unit of measurement in AI visibility analytics.

A reference to a specific source provided by an AI model alongside its response. Perplexity and Bing Copilot regularly cite sources. Being cited by an AI model drives direct referral traffic and is a strong signal of AI visibility.

A type of AI model trained on vast amounts of text data to understand and generate human language. Examples include GPT-4o (OpenAI), Claude (Anthropic), and Gemini (Google). LLMs are the technology behind AI assistants and chatbots.

The text corpus used to train an AI language model. LLMs learn brand associations, facts, and recommendations from their training data. Brands that appear frequently in high-quality training data sources tend to have better AI visibility.

A proposed standard file (similar to robots.txt) that websites place at their root to help LLMs understand their content structure. An llms.txt file provides a concise summary of a website's purpose, key pages, and permissions for AI training and retrieval.

A web crawler operated by an AI company to index web content for training or retrieval augmented generation (RAG). Major AI crawlers include GPTBot (OpenAI), ClaudeBot (Anthropic), and Google-Extended (Google). Websites can control AI crawler access via robots.txt.

A technique where an AI model retrieves relevant information from an external knowledge base before generating a response. RAG allows AI models to access up-to-date information beyond their training cutoff, and is used by Perplexity and Bing Copilot.

A text input given to an AI model to elicit a response. In AI visibility tracking, prompts are carefully crafted questions or requests that simulate how real users ask AI about brands, products, and categories.

The process of connecting AI outputs to verifiable, real-world sources. When an AI model is 'grounded', its responses are tied to specific documents or web sources rather than purely synthesized from training data.