September 25, 2014 Leave a comment
Each point of sale (POS) in a retail supply chain generates a goldmine of demand data. The data can be used to drive upstream decisions, but the amount of time, effort, and cross-team collaboration needed often frustrates such applications.
The MIT CTL researchers developed models to show how a leading CPG manufacturer can use large volumes of POS data to improve supply chain performance.
The thesis sponsor company, General Mills, Inc (GMI), is a Fortune 500 manufacturer of food products. GMI typically ships to the warehouses of large retailer customers via regional distribution centers. Of particular interest was finding out how POS data can be used to adjust production planning in order to reduce both production and inventory costs while maintaining high levels of service.
Four SKUs produced in the same manufacturing process and supplied to the same retail customers were selected for the purposes of the analysis. The SKUs represent a specific production platform, and as such, provide a good subject for testing the usability and added value of POS data.
To illustrate the value of this data source, the researchers focused on the potential for reducing two key manufacturing costs: change over and inventory holding costs. The research looked at how POS data integration in the supply planning process could produce direct benefits in terms of these costs while maintaining item fill rate targets set by the company.
The researchers designed a multi-period production planning linear program to optimize production scheduling for a given set of weeks. The program minimized total relevant costs subject to capacity and inventory target constraints. An important assumption was that all SKUs were being produced in the same plant. This allowed the linear programming to assign each SKU production quantity to each week for a unique factory location.
Three models were developed.
- The base model used only historical customer order data to plan production schedules.
- Model number two used POS data to forecast orders and adjust production to fulfill customer orders.
- The third model used POS data to adjust production to fulfill future POS demand, eliminating customer orders as an input.
Of the three models, the base option proved to be the most costly. Although the second one performed better, the improvement was relatively modest because this model fulfills according to customer orders and hence does not reduce the bullwhip effect. The third model, which uses POS data to improve demand forecasts and fulfill future demand, delivered the highest cost savings.
The findings suggest that the practical application of POS data can raise supply chain performance – with some riders.
First, companies gain the most benefit from applying POS data in this way when the bullwhip effect is minimal.
Also, in general, as the bullwhip effect increases so do inventory volumes and levels of stress on the production system, and the case for using POS data to offset these effects becomes stronger. However, in such situations manufacturers need to persuade their retailer customers to place orders that are aligned with the POS data, and collaborate with them to address misalignments.
POS data can also be used to alert manufacturers that they need to adjust customer order volumes with respect to actual sales.
Perhaps the most significant lesson is the importance of effective communications between manufacturers and retailers in realizing the value of POS data. The prize is well worth the effort – stronger long-term relationships that enhance competitive advantage.
This article was originally published by Supply Chain Management Review