Abstract

Organizations increasingly depend on networks of representations through which intent, decisions, work, evidence, outcomes, learning, and governance become available for coordinated action. Prior work on Organizational Correspondence argues that organizations preserve cognitive coherence when meaningful relationships among such representations are maintained over time. Yet organizational measurement remains organized around entities, outcomes, systems, processes, traces, or scores rather than the preserved relationships among representations. This paper develops a conceptual measurement framework for Organizational Correspondence in AI-native organizations. Its central contribution is to shift organizational measurement from isolated objects and outputs toward representation-network relationships while preserving the validity conditions that make such measurement interpretable. Drawing on measurement theory, construct validation, multi-level theory, temporal methods, organizational measurement, information quality, traceability, network analysis, composite-indicator methodology, and responsible AI governance, the paper specifies what is measured, at which levels, with what evidence, under which uncertainty conditions, and for which permitted uses. The framework distinguishes measurement from assessment, evaluation, diagnosis, and recommendation; defines a Correspondence Measurement Instance as the bounded record of a measurement execution; and argues that profiles should precede aggregation. It contributes to organization theory by making representation-network correspondence empirically tractable while preserving its relational, temporal, and contested character. It contributes methodologically by specifying validity, reliability, traceability, comparability, and claim-permission requirements for future empirical studies. The paper concludes with a staged research agenda for validating Correspondence Measurement in AI-native organizational contexts.