A variety of models have emerged in the field of energy system analysis to answer a wide range of research questions centred around a sustainable future for the energy sector. Even models designed to address similar issues often have a different focus or modelling approach. Thus, model experiments are a vital tool to provide an overview of the range of models and enable decision-makers to make meaningful model choices. Such comparisons are executed based on a harmonised data set to ensure a high degree of comparability. In the MODEX project cluster, six model experiments, including 40 energy system models, were conducted, and efforts were made to harmonise the input data within the individual comparison and beyond them in the consortium. The experiences and findings of the consortium on how data harmonisation could be performed are presented in this paper. In particular, the focus lies on data transparency to ensure a high degree of reproducibility. A key finding is that while model heterogeneity complicates harmonisation, an early focus on data research and scenario design promotes the creation of a common data set. The metadata collection can provide a significant advantage for the use of model experiment results by external scientists and the data acquisition process itself because of the predefined machine-readable and standardised format.