This study examines production losses caused by dropped components in a semi-automated automotive assembly process and evaluates a simple, data-supported supervision approach to reduce these events. One year of loss records was combined with structured Root Cause Analysis and on-site observations to understand when and why drops occur, linking quantitative patterns with specific ergonomic conditions and operator routines. Based on these findings, several targeted improvements were introduced, including adjustments to part feeders, clearer handling standards, small ergonomic changes and a lightweight dashboard showing drop-related indicators. After implementation, the assembly lines recorded a clear reduction in both the number and financial impact of drop events. The results show that these losses follow repeatable behavioural and task-related mechanisms and can be reduced through focused, low-cost measures. The study also shows that drop losses represent a form of operator-dependent waste that is not captured in aggregated performance metrics. By combining simple supervisory tools with established analytical methods, the proposed approach offers a practical way to improve stability in manual and hybrid assembly environments without requiring extensive digital infrastructure.