Demand sensing greatly improves forecast accuracy

Demand sensing software can dramatically improve companies' forecasts of the amount of goods they need to place on retail shelves, according to "Forecasting Performance Benchmarking Study—2011," a new report from supply chain software vendor Terra Technology. For the report, the software firm reviewed more than 7 billion physical cases of product, representing more than $200 billion in sales in 2009 and 2010 for nine leading consumer packaged goods (CPG) manufacturers. Those companies account for more than one-third of all CPG sales in North America.

Unlike traditional forecasting, which relies on historical sales data, demand sensing is a forecasting method that batches daily point-of-sales data and other relevant information, such as inventory levels, in its mathematical calculations to forecast demand for products. The study found that the demand sensing approach reduced the manufacturers' average weekly forecast errors by 40 percent for the period 2009 to 2010.

New products, which have no history, were more difficult to forecast than existing items that have historical sales data. Companies that use demand planning for their forecasting had an average weekly error rate of 65 percent on new product introductions. However, those that used demand sensing for new product forecasting saw an average weekly error rate of just 44 percent over a period of one year. It should be noted that companies that used demand sensing experienced higher error rates in the first few weeks after a new product introduction, but those rates dropped significantly by the 11th week.

The nine CPG companies that participated in the study were Campbell Soup, ConAgra Foods, General Mills, Kimberly-Clark, Kraft Foods, Procter & Gamble, S.C. Johnson, J.M. Smucker, and Unilever. Terra Technology said that the report encompassed virtually all items moving through all of those companies' North American warehouses.

For more about the study's findings, go here.