An ambidextrous perspective on business intelligence and analytics support in decision processes: Insights from a multiple case study

a r t i c l e i n f o Providing data-centric decision support for organizational decision processes is a crucial but challenging task. Business intelligence and analytics (BI&A) equips analytics experts with the technological capabilities to support decision processes with reliable information and analytic insights, thus potentially raising the quality of managerial decision making. However, the very nature of organizational decision processes imposes conflicting task requirements regarding adaptability and rigor. This research proposes ambidexterity as a theoretical lens to investigate data-centric decision support. Based on an in-depth multiple case study of BI&A-supported decision processes, we identify and discuss tensions that arise from the conflicting task requirements and that pose a challenge for effective BI&A support. We also provide insights into tactics for managing these tensions and thus achieving ambidexterity. Additionally, we shed light on the relationship between ambidexterity and decision quality. Integrating the empirical findings from this research , we propose a theory of ambidexterity in decision support, which explains how such ambidexterity can be facilitated and how it affects decision outcomes. Finally, we discuss the study's implications for theory and practice. Data-centric decision support is vital for managerial decision making in organizational decision processes. Business intelligence and analytics (BI&A) equips analytics experts (i.e., analysts or data scientists) with the technological capabilities to support decision processes with reliable information and analytic insights [1–4]. The added value of BI&A is based on increasing the utilization of " data-driven " decision making and thus improving decision quality and organizational performance [5–7]. However , realization of these benefits is not assured, and the very nature of organizational decision processes poses challenges for effective BI&A support. First, the reality of organizational decision processes has often been characterized as nonroutine and ill-structured [8–11]. In these situations , ambiguity prevails and the right questions are not always obvious at the outset. Rather, questions and solution alternatives are developed as part of the decision process and are subject to change [8,10]. As a consequence , data processing and analytics requirements can change frequently [12]. To achieve effective decision support in such nonroutine processes, the analysts who are involved must be able to adjust to these changes and, as a consequence, must maintain a high degree of adaptability and flexibility in their procedures. Second, effective decision support with BI&A requires analysts to have a high level of specialization in analytics, which is different from the domain …