Forecasting the Slow Movers

JDA Software Group is set to roll out this quarter a new forecasting algorithm that will address slow and erratic items for its E3 replenishment suite.

David Johnston, JDA’s senior vice president of manufacturing and wholesale distribution, said the algorithm, announced Feb. 13, will solve the challenges many companies face in using traditional methods to try to forecast slow movers.

“The algorithm takes into account that these items are erratic and you can’t forecast them in the same way you would forecast an item that sells at a good pace,” Johnston said.

He defined slow movers as products that sell less than one item per week at retail level. He said they can represent 60 to 90 percent of a retailer’s assortment. The algorithm responds to detected changes in sales patterns and considers influencing factors such as seasonality to perform dynamic exponential smoothing of forecasting statistics, Johnston said.

“When a sale occurs, the algorithm considers how long it has been since the item last sold and determines the fractional rate of sale normalized from the last sale,” he said. “Traditionally, you would only look at sales against forecasted sales for that period.”

Johnston said the algorithm will also adjust forecasts based on recent changes in sales patterns and reduce exception alerts by smoothing certain factors, such as a SKU that normally sells one item in a given period suddenly selling two items, which is a minor variation but gets recorded as a 100 percent forecasting error.

“E3 [will be able to] truly highlight exceptions that need action without cluttering up exception lists,” he said.

Rick Amari, president of Columbus Consulting, said that slow and erratic movers represent a big forecasting problem for many retailers.

“If an item has an average sale of 0.1 per week, when do you fill it?” Amari said. “It’s a real hard nut to crack.”

He said his firm has retail clients that end up spreading items all over their chains rather than making a chainwide buy of them, which can cause forecasting and replenishment issues if those items become slow movers.

“You get a proliferation of items that don’t move because they’re not in many stores,” Amari said. “The stuff lasts forever. Anything you can do to get those items into the right places and out of the wrong places is beneficial.”

According to Johnston, the addition of the new algorithm for slow-moving items is one of a number of recent E3 enhancements, including added functionalities for consolidated buying, multitier forecasting and replenishment, multilevel demand management, and configurable rules for top-down and bottom-up forecast reconciliation that allows lift predictions based on characteristics of a specific event or promotion.