Transmission, signification, decision. A three-level model for governing organisational information

Organisations record work with increasing precision and continue to make poor decisions. Data governance, business process management and organisational-routine research each explain part of this failure, but their concepts do not provide a common way to locate the point at which information ceases to support responsible action. This essay proposes a three-level model derived from Warren Weaver's distinction between the technical, semantic and effectiveness problems of communication. In organisational information governance, the three levels are transmission, signification and decision. Transmission concerns the recording and transfer of a sign, together with its provenance and integrity. Signification concerns the capacity of different parties to reconstruct compatible meanings from the same sign. Decision concerns the choice of what must change, who must behave differently and what cost that change imposes. The model does not claim that the three levels form a hierarchy of organisational maturity. Each can fail while the others appear to function. The argument is theoretical and is illustrated through two anonymised professional cases: a document lifecycle in a regulated financial organisation and a sequence of daily meetings in a digital-transformation project. The first case shows how a control system can establish provenance while leaving interpretation to professional judgement. The second shows how a meeting can repeat information already recorded because the organisation has no shared procedure for turning observation into a decision. The essay introduces shared interpretive capacity as a condition of signification, distinguishes a change in representation from a change in work, and reads the human-oversight requirements of Article 14 of Regulation (EU) 2024/1689 as a demand for competence, interpretation, contestation and the power to stop a system. It concludes with a diagnostic criterion in five questions: provenance, signification, contestability, measurement, and behaviour and cost. The model remains a working hypothesis without empirical validation.