The Message Transformation Model (MTM), for modeling complex message transformation processes in data centric application scenarios, provides strong capabilities for describing the data and control flow, transactional behavior and even the interaction with external systems. Thus, this general model could be used for different integration platforms like EAI-Servers, Message-Brokers and Subscription-Systems, as well as for ETL-Tools. In this paper, we describe self-optimization strategies for MTM processes to determine an optimal executable process. Our proposed strategies can be distinguished into rule-based and workload-based techniques. Aside from theoretical consideration, we describe implementation aspects within the integration platform TransConnectR . Furthermore, we present some first evaluation results.