Scenario: Stockout avoidance optimization
IBM® Sterling Intelligent Promising allows you to optimize stockout avoidance with the help of AI-powered models. Your business goal is to keep just enough merchandise in stock to avoid being out of stock. You can create a parameter profile that optimizes stockout avoidance costs.
The following example demonstrates how the inventory model takes a potential stockout at a node into consideration and ultimately optimizes the order to avoid a stockout at that node. When a sale occurs at this node in the future, the benefit is claimed as the stockout avoidance benefit, which is then reported in Benefits report.
Example
In a scenario where an order can be shipped from multiple nodes and the Avoiding stockouts objective is prioritized, Optimization service considers a potential stockout to select a ship node. It then optimizes the order so that the initial investment results in larger savings from stockout avoidance.
For example, Optimization service might ship an order from a node that is further away from a destination to avoid a stockout at a node that is closer to the destination. As a result, you can reinvest part of your stockout avoidance benefit toward the additional shipping cost. If the store that avoided the stockout sells that item to a walk-in customer, the benefit that is obtained by avoiding the stockout might be higher than the initial shipping cost.
| Shipping node | Distance between node and destination | Shipping cost | Processing cost | Load-balancing cost | Distance penalty | Stockout avoidance cost | Markdown avoidance cost | Total cost of optimization with inventory model | Total cost of optimization without inventory model |
|---|---|---|---|---|---|---|---|---|---|
| Node A | 759 miles | $4.93 | $3.50 | $.007 | $.0023 | $11.18 | $0 | $19.6186 | $8.4392 |
| Node B | 352 miles | $4.44 | $3.25 | $.003 | $.0012 | $77.41 | $0 | $85.1012 | $7.6946 |
- Total cost of optimization with inventory model = Shipping cost + Processing cost + Load-balancing cost + Distance penalty + Stockout avoidance cost - Markdown avoidance cost
- Total cost of optimization without inventory model = Shipping cost + Processing cost + Load-balancing cost + Distance penalty
If the inventory model was not used in this example, node B would likely be selected to fulfill the order since it is closer to the destination. However, the inventory model predicts that if the order is shipped from node B, a high possibility of a stockout at node B exists in the future. The predicted loss due to stockout at node B is $77.41 compared to $11.18 at node A. To minimize the loss due to stockout at node B, node A is selected for shipping. As a result, the total cost of optimization at node A is $19.6186 compared to $85.1012 at node B.