How to read our AI scores and insights.
Understand the limitations of AI visibility scores and how Zaher.AI measurements should be interpreted. Last updated: December 20, 2025.
AI Output Variability
Zaher.AI uses advanced artificial intelligence models to analyze content and generate insights. Like all AI systems, our technology may occasionally produce outputs that are unexpected, incomplete, or require verification.
How GEO Scores Are Calculated
- Query sampling — each analysis run uses a curated set of queries relevant to your industry and target keywords. Not all possible queries are covered; query sets are updated periodically.
- Response parsing — we analyze the text and citations in AI-generated responses to determine presence, prominence, and sentiment of brand mentions. Responses vary naturally between runs.
- Score calculation — scores aggregate results across engines and query types, weighted by estimated traffic and relevance. Because AI engines update continuously, the same query may yield different responses over time.
Margins of Error
Every score and metric provided by Zaher.AI includes an inherent margin of error. This is standard practice in data analysis and reflects the reality that measuring complex phenomena like AI visibility involves uncertainty.
Probabilistic, Not Deterministic — Comparative, Not Absolute
Our analysis is probabilistic: results represent likely outcomes based on available data rather than guaranteed facts, and running the same analysis multiple times may produce slightly different results. Scores are most valuable when compared over time and against competitors, not read as absolute measures.
Third-Party AI Engines
Zaher.AI analyzes outputs from third-party AI systems that we do not own or control, including ChatGPT (OpenAI), Gemini (Google), Perplexity, Claude (Anthropic), and others.