The manufacturing industry is undergoing a rapid digital transformation in which data‑driven methods, automation, and sensor technologies play an increasingly central role. Although many companies aim to improve productivity, reduce variation, and make more informed decisions, these ambitions require access to detailed and reliable production data. In practice, however, many organizations lack structured approaches for collecting and utilizing such data, resulting in limited insight into how time is distributed within production flows. When real production data is unavailable due to confidentiality, restricted access to industrial environments, or outdated technical infrastructure, simulated production systems can serve as an alternative. Digital models enable the generation of structured, repeatable, and realistic data that reflect industrial conditions, making it possible to study cycle times, time structures, and flow behavior without relying on physical measurements. This thesis investigates how different levels of data collection in a simulated production environment affect the ability to analyze cycle times and identify time‑critical steps in a production flow. By comparing a process without sensors to a process equipped with virtual conveyor sensors, the study examines how increased observability influences the quality and depth of analysis. The results show that detailed sensor data provides significantly better conditions for systematic improvement work, while manual analysis remains possible when sensors are not available. Furthermore, the study demonstrates how simulated sensor data can be used to identify bottlenecks and support production efficiency improvements before changes are implemented in real systems.