In the recent past new technologies have made a major contribution to improve the performance of athletics. Nevertheless, conventional solutions for athlete training development and analysis mainly rely on physical performance metrics, such as the amount of covered tracks and durations. Little has been done to monitor the quality of the physical motion and actions. This limitation stems from lacking suitable qualitative metrics and measurement systems. This paper investigates a new approach towards human activity recognition for individual performance analysis, leveraging sensor data from Inertial Measurement Units (IMUs). We show how to integrate sensors into the seat post of a bicycle and that it is possible to derive human performance indicators. The results of this study encourage the further exploitation of our method to analyze how cyclists behave while riding.