Executive Briefings

Implementing Forecast Value-Added Analysis

NewellRubbermaid executives realized they needed a process to track the multiple forecasting models the company uses to be able to determine what efforts makes them better or worse. Forecast valued-added analysis is the answer, say Ryan Rickard and Sean Schubert, senior supply chain managers at the company.

Though there a number of different forecasting models, when all is said an done, forecast value-added analysis suggests that the simplest, the naïve forecast, is hard to beat, according to both Rickard and Schubert.

Inputs to FVA starts with the naïve because it's easy to compute. Consensus demand forecasting, raw sales-order history data and statistical forecasting are additional inputs.

At NewellRubbermaid, forecast value-added analysis is built into the company's SAP backbone. Data and reports are used by several internal customers, including the demand planning community, which is always interested in improving its forecasting ability.

"They run the FVA reports and quite often sit down with sales and marketing and trace the evolution and see if there is room for improvement of their forecasts," Schubert says.

The naive forecast model is especially useful with sporadic or unpredictable demand. "The naive comes out pretty good," Rickard says. "The statistical modeling techniques we're using aren't as optimal as they could be. So if the statistical forecast is not beating the naive, it indicates that the stat forecast can be reviewed and adjusted."

To view video in its entirety, click here

NewellRubbermaid executives realized they needed a process to track the multiple forecasting models the company uses to be able to determine what efforts makes them better or worse. Forecast valued-added analysis is the answer, say Ryan Rickard and Sean Schubert, senior supply chain managers at the company.

Though there a number of different forecasting models, when all is said an done, forecast value-added analysis suggests that the simplest, the naïve forecast, is hard to beat, according to both Rickard and Schubert.

Inputs to FVA starts with the naïve because it's easy to compute. Consensus demand forecasting, raw sales-order history data and statistical forecasting are additional inputs.

At NewellRubbermaid, forecast value-added analysis is built into the company's SAP backbone. Data and reports are used by several internal customers, including the demand planning community, which is always interested in improving its forecasting ability.

"They run the FVA reports and quite often sit down with sales and marketing and trace the evolution and see if there is room for improvement of their forecasts," Schubert says.

The naive forecast model is especially useful with sporadic or unpredictable demand. "The naive comes out pretty good," Rickard says. "The statistical modeling techniques we're using aren't as optimal as they could be. So if the statistical forecast is not beating the naive, it indicates that the stat forecast can be reviewed and adjusted."

To view video in its entirety, click here