This year, ARC compared the features and functions of leading suppliers’ solutions for demand management. Based on this analysis, we created a functionality pyramid that shows basic (features that most suppliers have) to more advanced functionality (functions that relatively few vendors have, or that relatively few exhibit in a robust manner). Price optimization appeared at the top of the pyramid, followed by automated selection of optimal forecasting algorithms by stock keeping unit (SKU), lost sales analysis and demand decomposition capabilities.From a functionality perspective, demand solutions have noticeably improved in the past few years in providing support for sales-and-operations planning (S&OP), a key supply chain process. Some level of S&OP analytics can be embedded in either demand or supply planning, although some vendors offer separate analytics modules to support S&OP. Not all vendors’ solutions do a good job of documenting the assumptions, risks and opportunities discussed during the S&OP meetings on which planned actions are based.Demand decomposition functionality can improve the forecasting performance. In many instances, the historical demand pattern can be somewhat misleading due to external factors such as competitor price breaks, promotions and other marketing initiatives. Demand decomposition segments the demand history into base level demand, based on historical time series data, and the impact of marketing and special events on the demand for a stock keeping unit (SKU).Automation to help with selecting the optimal forecasting algorithm is also important. However, demand is a moving target. Over time, demand patterns can change, and the forecasting algorithms need to change as well. Several years ago, demand applications became self-grading. They offered alerts that, in effect, said, “I’m not performing well.”We are in our third generation of self-grading functionality. In the first generation, the application created automated alerts when the forecast error passed a preset threshold. Users then needed to manually figure out what algorithm would work better for a certain SKU.In the second generation, the application had automated alerts, but also automated the selection of the new algorithm by running multiple models against the demand history. However, there were unintended consequences to this automation. This functionality proved to be too “noisy” in many instances. The solution could end up flip-flopping on the best algorithm for a particular SKU many times a year.
Guided decision makingThe third generation solution has automated forecast error alerts, but is actually less automated when it comes to selecting an algorithm. Instead, it might be described as “guided decision making.” Third-generation solutions help forecasters ensure that input streams are accurate by decomposing the history streams before feeding them into the “pick best” engine. They then take the baseline demand and classify it by what type of demand pattern it exhibits (seasonal, continuous, erratic and the like). Based on the demand pattern, a best-fit algorithm is applied, which then needs to be tuned.Price optimization is a logical extension to what a robust demand planning application does. However, pricing optimization solutions are usually bought and used by people within the sales and marketing function, and at this point, there is little or no collaboration with supply chain personnel to shape demand. Currently, within the S&OP process, most companies attempt to shift supply to fit demand. Price optimization would provide the alternative capability to shift demand to fit supply. Further, this is a lever that would improve profitability. Eventually, demand shaping will come to S&OP, but we are clearly in the early adopter stage now.
Steve Banker,
[email protected], is Director, Supply Chain Management, at ARC Advisory Group Inc., in Dedham, Mass.
Subscribe to Automation World's RSS Feeds for Columns & Departments
