Multi-stage tools are among the most economical forming tools but due to the high development costs, their use only pays off in large-scale production. Accelerated commissioning offers great potential for maximizing economic efficiency. By integrating sensor technology into tools, knowledge can be extracted from the process using data-driven methods to identify problems. Extensive process data is usually collected for this purpose. Commissioning with limited process data requires data-efficient methods and the use of synthetic data for fault identification during commissioning. However, the combination of synthetic and real data poses an additional challenge. This paper presents a methodology to identify deviations between synthetic and real data and their causes.