- Data Analytics
- Demand Planning
Order fill rate is a critical performance metric in retail supply chain operations. Retailers use it to ensure deliveries received from their suppliers are in full quantities as per order. The retailers levy fines on suppliers that fail to comply with the metric. C.H. Robinson has a division that provides retail consolidation services to multiple suppliers. It arranges consigned inventory from multiple suppliers, stores it, and ships it to retailers in full truckloads as per order demand. They are interested in designing an inventory strategy that ensures a 98% order fill rate, thereby minimizing fines charged by retailers. An inventory strategy is focused on three key aspects i.e., optimal review interval, order quantity, and safety stock requirement. This project uses historical order and inventory data provided by C.H. Robinson to design an inventory strategy. The methodology taken is to narrow the focus down to 50 top-selling SKUs out of a total of 3,769 that consistently represent a significant share of the total shipments out of the distribution center. Upon identification of top-selling SKUs, two steps are taken to build a strong foundation before creating an inventory strategy. A forecast is built using techniques such as autoregressive integrated moving average (ARIMA) and error trend and seasonality (ETS) to ascertain the historical volatility in demand. After which the research uses the forecast accuracy to build optimal inventory levels required to achieve order fill rate targets. Furthermore, SKUs that show similar characteristics in terms of fill rate, volatility, and forecast accuracy are segmented into three clusters using k-means clustering. Thereafter, a periodic review inventory control system is used to obtain the optimal review intervals, order quantity, and safety stock levels for each of the three clusters. The research paper suggests an optimal amount of inventory that C.H. Robinson should hold in its DC to ensure an order fill rate of 98%. It also compares it with existing inventory levels maintained at the DC for each cluster, and the corresponding fill rate performance for each cluster. Ultimately, the research paper explores the trade-off of higher inventory holding costs associated with maintaining inventory levels geared towards achieving a 98% order fill rate performance. The research paper also provides C.H. Robinson with a framework they can use to make the best financial decision, given the trade-off mentioned above.