Human workers face major challenges with the increasing complexity in modern production environments. On the shop floor, a wide variety of assisting systems support them in their daily tasks. However, this requires profound knowledge of the underlying tasks and processes to implement highly effective assistive systems which are well accepted by the workforce. This paper describes an approach for analyzing assembly tasks executed by human workers. The analysis combines a manual approach using video and an automatic approach based on sensor data. Further, we use the General Assembly Task Model (GATM) to structure assembly tasks in four phases. This helps us to gain a deeper understanding of manual tasks and it further helps to identify in which situations of an assembly process the workers need assistance and choose the right measure to support them.