Next Generation Demand Management

Chapter 6The Analytics

If the demand planners (or demand analysts) at your company are using advanced statistical models like ARIMA, ARIMAX, and dynamic regression using causal factors (i.e., price, intervention variables to capture sales promotion lifts, and outliers, along with other causal factors), then you are among a handful of companies that are doing true demand-driven forecasting and planning using advanced statistical modeling. It is amazing that in 2016, according to an Industry Week study, 77 percent of demand forecasters still use Excel, and the number one mathematical method being deployed is moving averaging.¹ This is embarrassing to say the least for demand forecasting and planning. With all the data collection, processing, and technology advancements over the past two decades, companies would rather use buffer inventory stock to protect against demand variability than invest in new skills, analytics, and technology to improve their demand management process. In fact, companies are still cleansing their demand history manually to separate baseline and promoted volume due to the restrictions of their legacy ERP systems as a result of the limited array of statistical methods (moving averaging and exponential smoothing methods) available in their model repositories. Those methods can only model patterns associated with trend, seasonality, and cycles.

As we discussed in Chapter 3, companies can holistically model the baseline trend, seasonality, correct for outliers, ...