Improving delivery performance is a critical requirement for distribution operations, as delays often arise from documentation errors, non-standardized procedures, and irregular demand patterns. This study proposes an integrated solution to enhance the On-Time Delivery (OTD) indicator in a logistics process by applying three complementary engineering tools: work standardization, poka-yoke mechanisms, and a multi-SKU forecasting model. Time measurements were conducted using a representative sample of 370 orders, allowing the identification of fixed and variable components in the generation of shipping guides and their optimization under a standardized workflow. Additionally, historical demand was analyzed to develop a forecasting model capable of managing highly irregular consumption, enabling better inventory availability. A simulation of 100 replications was performed to validate the improvements and evaluate their stability under operational variability. The results demonstrate a reduction in processing times, a decrease in documentation errors, and a more stable performance distribution, contributing to an overall improvement in OTD. Limitations related to data variability and manual dependencies were identified, and potential future enhancements, such as automated verification and real-time tracking, are suggested.