Manufacturing companies in machinery and plant engineering face a variety of challenges. Due to the conditions of the VUCA world (volatility, uncertainty, complexity, ambiguity) and increasing customer requirements, manufacturing companies must constantly review and adapt planning decisions or even implement new ones. This need for continuous adaptation especially applies to decisions in production planning. Challenging situations such as the COVID-19 crisis or the war against Ukraine are particularly demanding for production planning and require labor-intensive, manual intervention in planning. Additionally, these re-planning activities must be carried out in a short period of time. Decision-making based on data-driven approaches, like machine learning (ML) or simulation, will facilitate these rapid adjustments under tight time constraints. Not only can data-driven approaches predict how a customer-specific order will behave during the processing of the order in the company, but planning decisions can also be simulated and evaluated. This publication presents the concept of a methodology with three solution modules. A method for predicting order behavior based on machine learning is provided as the first module. The second module consists of a method for interconnected simulation objects that make it possible to model planning decisions. The last module is used to find an optimal planning decision.