Finding the Root Cause of Forecast Inaccuracy
Demand forecast accuracy is the foundation of effective supply planning. Without it, which we like to call forecasting inaccuracy, all subsequent efforts to optimize plans and policies will miss the mark, no matter how well-executed.
When selecting supply planning software, it is crucial to understand the importance of forecasting relative to other system features. It is common to focus on addressing issues such as transferring stock between warehouses to mitigate imbalances, prioritizing customer orders in a stock-out situation, managing expedited deliveries, squeezing more stock into limited warehouse space, etc. While these activities are important, they are not the root cause of the problem, they are artifacts of inaccurate forecasts (the true underlying problem). Get the forecast wrong, and it doesn’t matter how well you execute all of the downstream activities; you will still encounter stock outs, bloated inventories, product obsolescence, and a host of other ills.
Vanguard eliminates forecasting inaccuracy problems by applying advanced analytics that addresses root causes, to improve efficiency through AI-driven automation, and facilitates collaboration through an easy-to-use, Web-based interface. This is very different from many competing solutions, which focus on improving planner efficiency simply by giving them easier access to data and reports without materially changing planner activities.
Vanguard’s process starts by getting the demand forecasts right through a blend of predictive analytics and human insight. A black-box forecast engine will never produce optimal forecasts. Neither will a human-driven process based on manual forecasts.
Subtle differences in how a system facilitates collaboration between users or merges insights with statistics can make a dramatic difference in forecast accuracy.
It is common to refer to a single point estimate as a forecast (for example, “we expect to sell 100 units next month”). However, a forecast is really a continuum of outcomes where the point estimate is only the average. For example, a forecast of 100 units +/-10 is very different from a forecast of 100 units +/-50. While both forecasts have the same point value (100), the latter forecast requires more safety stock to maintain a profit-optimized stocking level and results in very different production and inventory plans.
Supply planning applications typically separate the demand planning and supply planning activities. In fact, most systems require planners to prepare forecasts outside the planning system using spreadsheets and then upload them to support an MRP or similar scheduling process. Systems are designed this way because forecasting is difficult to do correctly.
Following are problems with this approach:
• The lack of accuracy and timeliness associated with a spreadsheet-based process
• A forecast is not a single point estimate, but a continuum, making it nearly impossible to manage forecasts in a spreadsheet or to upload the results
Vanguard tightly integrates the demand and supply planning functions, letting you consider all possible outcomes, not just the point forecast, when creating a profit- optimized supply plan. This translates into reduced safety stock and improved service levels.