Continuous improvement requires visualizing process constraints which interrupt workflows. Production control from a management perspective often operates at lower levels of information granularity than required at operational levels to perform work without interruptions. If not controlled in detail, causes and effects of workflow interruptions remain unclear in environments of high complexity and non-standardized work.Workflow efficiency has been studied through work sampling or time-motion studies, estimating shares of direct work. However, few studies exist that show how to create digital representations of workflows and analyse them for interruptions, contributing to smoother workflows. The paper examines workflows of plumbing work from video footage. This video data is classified and analysed for frequency, causes, and effects of work interruptions.Results indicate that value-supporting activities caused the largest proportion of interruptions. Moreover, the proportion of non-value-adding activities increases when durations of interruptions rise. Based on the results, the paper contributes to further understanding of workflow interruptions in plumbing work. Finally, it provides suggestions on how to close gaps of information granularity between management and operational levels, through the development of simulation models and the application of automated data collection, contributing to developing digital twins of construction processes.