AI Search Brand Visibility Measurement Protocol: A Reproducible 100-Query Research Method

This working paper presents a reproducible 100-query research protocol for measuring brand visibility across AI-assisted search systems. It treats brand mentions, explicit recommendations, source citations, clickable links, list position, and response stability as separate outcomes rather than combining them into an unsupported “AI ranking score.” The protocol uses 100 predefined queries divided into five intent categories, three repetitions per query, and three non-consecutive observation windows. It provides detailed query-selection criteria, observation controls, coding rules, metric formulas, data fields, inter-coder agreement procedures, evidence-retention principles, and minimum reporting standards. The method is designed to support transparent cross-platform and longitudinal comparisons while accounting for response variability, language, location, personalization, retrieval conditions, and product changes. It does not claim access to proprietary algorithms, model training data, hidden ranking signals, or official platform metrics. The protocol does not guarantee brand mentions, recommendations, citations, referral traffic, rankings, or commercial outcomes. It is intended as an independent working method for researchers, digital visibility professionals, organizations, and practitioners studying AI-assisted discovery.