MIT Research: How to Cut Costs by Improving Product Punctuality

Editor’s Note: Brad Gilligan and Huiping Jin, created the model for their MIT Supply Chain Management Program master’s thesis. The work was carried out for a major US retailer, and the project was supervised by MIT CTL Research Director Dr. Edgar Blanco. For more information on the program, visit http://scm.mit.edu/program

Products received from suppliers are not always available exactly when retail channels need them. Items that arrive way too early clog up warehouse space while latecomers often incur expediting costs and lost sales. By tailoring transportation methods and distribution center processes to match the delivery speed required of each product, companies can reduce the cost of transportation, prevent excess inventory, and eliminate lost sales.
Configuring supply chains in this way is often done through SKU segmentation, but this is difficult in complex operations where there are numerous products and an extremely diverse supplier base.

Researchers at the MIT Center for Transportation & Logistics (MIT CTL) have developed a model that uses purchase order (PO) information to help retailers determine when many different types of products need to be shipped to meet sales deadlines.

Lack of visibility

This retailer excels at capturing opportunistic business. For example, it might buy a product at low cost in the off season for sale during periods of peak demand.

This type of operation usually involves suppliers and stores in multiple countries. In order to make the project more manageable, the research focused on items purchased in China for sale in the United States. Even with this limitation the analysis covers more than 30,000 POs, 40,000 unique SKUs, and 1,000 suppliers.

Some products were received well in advance of the sale date, while others arrived with very little time to make the final delivery to the store. Moreover, the company was unable to see which products fell into the early and late categories.

There was a huge opportunity to achieve savings by introducing slower, more cost-effective supply chains for early items and faster delivery times for tardy products. The model aimed to achieve this by adjusting lead times so that zero to seven days elapsed between a product’s arrival at the DC and its required availability at the store.

Performance improvements

The researchers were able to identify PO attributes that can be used to predict with more than 90% accuracy how much time the retailer needs to ship a product from suppliers. In test simulations on-time performance was improved by 36% to 60%. Additionally, no products arrived late and those that were received before the agreed delivery date were less than 10 days early.

Using this information, the retailer can cut transportation and holding costs, and potentially increase total sales volumes because more products will be available at the store when required. Also, the model can be updated automatically as new data comes in.

Although the model is geared to the needs of one US retailer, it is of interest to any retail operation that procures product opportunistically.

SC
MR