Glossary

Key terms and definitions used in AI perception management. Reference guide for understanding VectorGap metrics and concepts.

Core Concepts

AI Perception: How language models understand, represent, and discuss a brand, product, or organization. Encompasses factual accuracy, sentiment, visibility, and recommendation likelihood.
Generative Engine Optimization (GEO): The practice of optimizing content so that AI systems can understand, retrieve, and cite it accurately. The AI-era equivalent of SEO.
Hallucination: When an AI generates false or fabricated information with high confidence. In brand contexts, this includes wrong pricing, non-existent features, or mixing up companies.
Entity Confusion: When AI confuses your brand with another company, typically one with a similar name or in the same industry. A common source of hallucinations.
Extractability: How easily AI systems can extract and cite information from your content. Determined by format (lists, tables), structure (answer-first), and freshness.
Matthew Effect: In AI citations, the phenomenon where already-visible brands get more recommendations, while unknown brands remain invisible. Based on the biblical "rich get richer" principle.
AI Overview: Google's AI-generated summary that appears at the top of search results. Synthesizes information from multiple sources and increasingly replaces traditional organic clicks.
Citation Authority: The likelihood that AI systems will cite your content as a source. Influenced by content format, freshness, original research, and brand search volume.
AI Crisis Management: The practice of identifying, responding to, and preventing AI-related reputation crises such as hallucinations, negative recommendations, or inaccurate information.
Local AI Optimization: Optimizing for AI visibility in location-based queries. Includes "near me" searches and local business recommendations in AI assistants.
Agentic Commerce: AI agents that make purchase decisions and transactions on behalf of users. Requires brands to optimize for machine-readable formats and APIs.
AEO (Answer/Agentic Engine Optimization): Optimizing content for AI assistants and agents (like Copilot, ChatGPT) so they can find, understand, and present answers effectively. Microsoft defines AEO as focused on driving clarity with enriched, real-time data. Distinct from GEO, which focuses on trust and credibility.Source: Microsoft Advertising (January 2026)
AI Shopping Ecosystem: Microsoft's framework describing three overlapping consumer touchpoints: AI browsers (interpret page context while browsing), AI assistants (answer questions and guide decisions), and AI agents (take actions like completing purchases). The touchpoint matters less than whether the system can access accurate, structured product information.Source: Microsoft Advertising (January 2026)
Discovery to Influence: Microsoft's framework describing the shift from SEO (discovery/clicks) to AEO/GEO (influence/recommendations). SEO helps products get found, AEO helps AI explain them clearly, GEO helps AI trust and recommend them.Source: Microsoft Advertising (January 2026)
Trust Signals: Indicators that AI systems use to assess content credibility: verified reviews, review volume, clear sentiment, press coverage, certifications, partnerships, and consistent structured data. Exaggerated or misleading claims reduce trust and limit AI visibility.Source: Microsoft Advertising (January 2026)
AI Browser: A browser enhanced with AI capabilities that interprets page content and surfaces contextual information while users browse. One of three AI shopping touchpoints defined by Microsoft alongside AI assistants and AI agents.Source: Microsoft Advertising (January 2026)

Metrics

Accuracy Score: A metric measuring what percentage of AI claims about your brand are factually correct when compared against your knowledge base.
Sentiment Score: A metric measuring the emotional tone AI uses when discussing your brand - positive, neutral, or negative.
Visibility Score: A metric measuring how prominently your brand appears in AI responses to relevant queries. First mention scores higher than later mentions.
Coverage Score: A metric measuring what percentage of your key features, differentiators, and use cases AI mentions when discussing your brand.
Credibility Score: A metric measuring how authoritative AI sounds when discussing your brand - whether it cites sources, uses confident language, etc.
Recommendation Score: A metric measuring how often AI actively recommends your product (not just mentions it) when answering relevant queries.
Share of Voice (SOV): The percentage of times your brand appears compared to competitors when AI answers category-level queries.
Brand Search Volume: How often people search for your brand by name. Research shows this has a 0.334 correlation with AI citations - stronger than backlinks for determining AI visibility.

VectorGap Features

Perception Audit: A systematic analysis of what multiple AI systems say about a brand. Involves querying ChatGPT, Claude, Gemini, and Perplexity and analyzing their responses.
VectorGap Perception Score (VPS): A composite score from 0-100 that measures overall AI perception quality. Combines accuracy, sentiment, visibility, coverage, credibility, and recommendation metrics.
Knowledge Base: A structured collection of verified facts about your brand - pricing, features, use cases, competitive positioning - that serves as ground truth for hallucination detection.

Technical

LLM (Large Language Model): The AI systems that power tools like ChatGPT, Claude, and Gemini. These models process and generate text based on patterns learned from training data.
llms.txt: An emerging standard for providing AI systems with structured information about a website. Similar in concept to robots.txt but for language model crawlers.
JSON-LD: A format for embedding structured data in web pages using JSON. Used by search engines and increasingly by AI systems to understand page content.
Schema.org: A collaborative vocabulary for structured data markup. Schemas like FAQ, HowTo, and Product help AI understand content meaning.
RAG (Retrieval-Augmented Generation): A technique where AI retrieves relevant content from external sources before generating responses. How tools like Perplexity provide cited answers.
Training Data: The corpus of text used to train language models. AI knowledge about your brand is limited by what exists in this data (often with a cutoff date).
Retrieval Layer: The search infrastructure that AI systems like ChatGPT use to find content before generating responses. Controls what information AI can access and cite.
Chunking: The process of dividing content into smaller segments (typically 200-500 words) that AI retrieval systems can index and retrieve. Self-contained chunks with clear answers perform best.
Crawled Data: Information AI systems learned during training and retrieve from indexed web pages. Shapes a brand's baseline perception including product categories, reputation, and market position. One of three data layers Microsoft identifies in AI recommendation systems.Source: Microsoft Advertising (January 2026)
Product Feeds & APIs: Structured data actively pushed to AI platforms, giving brands control over how products are represented in comparisons and recommendations. Provides accuracy, detail, and consistency. Microsoft identifies this as the second data layer in AI shopping.Source: Microsoft Advertising (January 2026)
Live Website Data: Real-time information AI agents see when visiting a site: rich media, user reviews, dynamic pricing, and transaction capabilities. The third data layer in Microsoft's AI shopping framework.Source: Microsoft Advertising (January 2026)
AI Recommendation Layer: The system through which AI assistants decide what to recommend. Combines crawled web data (brand positioning), product feed data (prices, specs, availability), and live website data (reviews, promotions, delivery). Brands compete for influence at this layer, not just search rankings.Source: Microsoft Advertising (January 2026)
Multi-modal Interpretation: The ability of AI systems to understand content across formats: text, images, video, and audio. Requires good alt text, video transcripts, and structured image metadata. Part of Microsoft's content optimization strategy for AI.Source: Microsoft Advertising (January 2026)

Quick reference

The 6 perception metrics

Accuracy

Are facts correct?

Sentiment

Is tone positive?

Visibility

Are you mentioned?

Coverage

Key features cited?

Credibility

Authoritative?

Recommendation

Actively suggested?