by Suresh Divakar, Brian T. Ratchford, and Venkatesh Shankar
This work won a finalist award for the 2004 ISMS-MSI Marketing Science Practice Prize.
http://techtv.mit.edu/collections/isms:1136/videos/7480-pepsi
Pepsico developed a model and decision support system called CHAN4CAST to forecast sales for consumer packaged goods. The model decomposes sales into several channels, including grocery, drug, convenience and gas, fountain and the like, decomposing sales forecasts into different channels. The model also incorporates holidays, temperature, trading day effects and new product introductions. The model’s mean percentage forecast error for the grocery channel in 2003 was -0.4%. The associated decision support system (DSS) allows product managers both to plan for forecast sales volumes as well as to simulate the effect of different firm and competitors’ actions on those sales. The tool includes a scorecard by account, channel and region that allows Pepsico to continually track its sales against forecast. The company estimates benefits as well over 1,000% return on the model investment.