This research seeks to broaden and strengthen the holistic understanding of IT governance effectiveness by specifically examining why IT governance systems often fail to produce appropriate or desired IT and organizational behaviours. To address this objective, we investigate and develop a theoretical framework for understanding and explaining the varied sources of divergence that occur during the enactment of IT governance mechanisms. Defined as the difference between desired behaviours and actual behaviours, we argue that the acceptance and consideration of all sources of divergence
within the enactment of IT governance mechanisms, is not only necessary, but critical to the appropriate design and maintenance of an effective IT governance system.

Traditional IT governance perspectives, heavily rooted in the structural and normative aspects of oversight and control (i.e. structures), have limited our ability to adequately and fully understand how IT governance performs in practice. Framing IT governance mechanisms as routines, we draw on institutional theory and organizational routines theory as an alternative lens for understanding why organizational behaviours are
not always aligned to those expected by IT governance owners. Based on Pentland and Feldman`s (2008) generative model of organizational routines, we establish a novel conceptualization for IT governance divergence that posits and delineates three primary sources of IT governance divergence: Representational Divergence, Translational Divergence and Performative Divergence.

Through the in-depth examination of the IT investment planning, prioritization and selection routines within two exploratory case studies, we inductively propose a model for explaining IT governance divergence. We
apply a narrative networks approach to frame and analyse qualitative data captured through semi-structured interviews, archival and document review and direct observation. Pattern-matching and emergent themes analysis is performed to identify and define first-order and second-order constructs, along with 15 relational propositions. Given the dearth of theoretically-grounded research in this domain, the central contribution of this study rests in the establishment of a robust theoretical framework of IT governance divergence upon which further cumulative empirical study can be undertaken.