<![CDATA[Background: Generative AI creates observable entity representations and citation environments. However, common visibility measures often rest on different measurement assumptions. Treating them as comparable can distort organizational diagnosis. Purpose: This study develops RADAR—Recognition, Authority, Discovery, Asset Citation, and Reach—as a governed diagnostic measurement specification for observable AI visibility. Design/Methodology/Approach: The specification organizes five parallel diagnostic domains. Each domain defines its measurement object, opportunity set, denominator, descriptive estimand, audit conditions, identity grain, and interpretation ceiling. Three purposively selected legacy audit datasets provide bounded demonstrations; they are not representative organizational samples. Findings: Owned citation-occurrence shares ranged from 1.6% to 11.8%, while 79%–89% of ownership classifications remained Unknown. HHI increased and the entropy-based Effective Number decreased when source identities were aggregated from pages to hostnames and registrable domains. Conclusion: Observable AI visibility is not a single self-interpreting quantity. Credible diagnosis requires each quantity to be reported with its measurement conditions and inference limits. Originality/Value: RADAR integrates heterogeneous focal-entity diagnostic quantities within one governed specification. It is not a validated universal index, causal model, or platform benchmark.]]>
Copyrights © 2026
