There's only one place where forecasting should be taking place in the supply chain today, says Martin: the retail store. "Everything else can be calculated," he adds.
Arriving at the correct numbers isn't easy, but it's not impossible. In a typical store, Martin says, 20 percent of products will sell daily and 20 percent weekly, leaving 60 percent that turn over with less frequency, and are traditionally tough to forecast accurately. "Integer forecasting" makes possible the quantification of demand for that difficult category, in a "time-phased" fashion. As sales occur, the retailer can calculate how its supply chain must perform in order to meet the balance of demand over a given period of time.
Traditionally, retailers might parse demand for slow movers based on average demand, often ending up with a decimal. "But people don't buy in decimals," Martin says. Hence the term integer forecasting. It involves the application of consumption logic, constantly re-forecasting and spreading unsold product over the balance of a specified period. The technique does a "decent job of forecasting 60 percent of products in a retail store," he claims.
Martin says retailers have been gradually warming up to the innovation, which has been "years in the making." The difficult part lies in convincing them that the calculation of demand patterns eliminates the need for forecasting in all places but the actual store. The idea "doesn't come across naturally to people," he acknowledges. "It's a foreign way of thinking how to manage product."
