The planning and control of modern logistics systems are characterized by significant complexity due to various dynamic influencing factors. Information systems are employed at multiple levels to support or even fulfill planning and control tasks. These information systems not only manage the material flow but also document material movements. This leads to a large amount of standardized process data. This article outlines how data-driven approaches within a holistic procedure model can be utilized to exploit this process data, to identify weaknesses, and to derive recommendations for action. Based on the identification of relevant input data and its aggregation to process information and key performance indicators, optimization potential can be identified automatically through a deviation analysis using a so-called system of interdependent effects. By applying a machine learning classification model, these potentials for optimization are assigned to defined root cause classes. Process experts can further analyze the identified root causes and make recommendations for action using supporting tools. The applicability and usability of the procedure model are validated through a case study.