Strategic governance of algorithmic decision systems in organizations

Purpose This study aims to identify and analyze the governance capabilities that enable organizations to strategically oversee algorithmic decision systems. The research examines how different governance mechanisms interact and influence each other within organizational settings. Design/methodology/approach A multi-stage research design was adopted. First, a systematic literature review was conducted to identify governance mechanisms associated with algorithmic decision systems. Second, an empirical survey was used to operationalize these mechanisms and validate their underlying structure using exploratory factor analysis (EFA). Finally, the fuzzy DEMATEL method was applied to examine the causal relationships among the validated governance capabilities and to distinguish between driving and dependent governance factors. Findings The results reveal that algorithmic governance operates through a set of interrelated organizational capabilities. Strategic oversight mechanisms, particularly board-level AI risk monitoring and AI governance committees, are key drivers of other governance practices. Procedural assurance mechanisms, such as algorithmic impact assessments and ethics-based auditing, play central roles in operationalizing governance oversight, while transparency, human oversight and organizational learning mechanisms function as dependent governance capabilities. Practical implications The findings provide guidance for managers seeking to design effective governance structures for algorithmic decision systems by emphasizing the importance of strategic oversight and institutional governance frameworks. Originality/value The study contributes by conceptualizing algorithmic governance as a sequenced capability system and empirically uncovering its causal structure.