The overall European Air Traffic Management aims for an efficient utilisation of ATM sub-systems like the European Air Traffic Management Network. Highly congested hub airports and controlled airspaces may lead to Demand-Capacity-Balancing issues during high demand periods. Air Traffic Flow Management as one of the European ATM domains resolves these issues with certain capacity balancing measures. Besides slot allocation, one frequently used measure is (lateral) re-routing, which often leads to less efficient trajectories resulting in additional CO2-emissions and costs for airspace users. Recent research suggests that the share of inefficient short-term re-routing measures could be reduced by high-quality predictions of future flight trajectories through Machine Learning methods. One essential requirement for the development of ML models is the optimal choice of input parameters. In this study, a structured analysis of an extensive set of flight plan data is conducted in order to identify key operational parameters for ATFM routing decisions. An extensive flight plan data sample covering three months in 2016 is parametrised and analysed to gather a sufficient data baseline. A set of representative Origin-Destination pairs with high demand rates are investigated in more detail. Results show that flights approaching large hub-airports have a higher chance of being re-routed resulting in a less efficient trajectory in terms of lateral ATS-efficiency. The parameters with the highest relevance on ATFM routing decisions were found to be lateral ATS-efficiency, demand along the individual sector profile as well as the weekday of departure for planned trajectories.