Supply chain planning leaders commonly struggle with forecast accuracy, or lack thereof, as they build planning capabilities for their businesses. Business functions outside supply chain often see forecast error as the disqualifier for the structured, formal planning process.
Forecast accuracy must be good enough before supply chain planning can be taken seriously. The cross-functional reluctance to buy in to supply chain initiatives because of forecast error can limit the company’s ability to improve the overall supply chain capabilities. The Gartner Supply Chain User Wants and Needs Survey (February 2019) shows the top internal obstacle for companies to achieve their supply chain goals is forecast accuracy and demand variability.
Stakeholders across the business often expect supply chain planning teams to fix the broken forecasting process by themselves. Professional demand planners can do a lot to improve the planning accuracy of the products in the product portfolio, but there is a limit.
The most commonly used forecast accuracy metric is Mean Absolute Percent Error (MAPE), which is measured at item level.
This means that for a forecast to be perfectly accurate, the prediction of the future actual sales volume of a specific SKU must be exactly what was actually sold. Also, the volume of a specific SKU that was actually sold must have been predicted in the past. Demand planners can analyze the forecast accuracy data once the measurement lag time has elapsed to find root causes and issue corrective actions. All this is well within the planning function’s scope of control.
One source of forecast error is largely outside of the demand planning team’s immediate control. The problem is the financial “plug.” The plug is a placeholder to fill a gap between a bottom-up, monetized forecast and the top-line revenue objective. The plug does not have item level detail. It is an undefined “bucket of monies” that balances financial plans to the business’ revenue goals. When a financial plug is in play, a portion of future demand is entirely unplanned as far as supply chain is concerned.
I have had several discussions about the financial plugs (i.e., placeholder, stretch goal, management override) with supply chain planning leaders, who usually agree that there is a difference between the supply chain plan and the finance estimate. The size of the actual placeholder varies by company. However large the gap is, the plug sets the maximum for the forecast accuracy.
For example, let’s say the monetized bottom-up demand forecast came to 80% of the revenue budget. If the supply chain demand plan and the financial forecast are aligned, the financial latest estimate then would have 80% of the revenue covered with a forecast with SKU level detail. The remaining 20% is entered at product family, region or even at business unit level, then communicated to the sales team as a “go-get” or “stretch goal,” or something similar. When the sales team succeeds in delivering to the stretch goal target, all of demand above and beyond the supply chain plan is forecast error. Therefore, a plan with 20% financial plug translates to maximum forecast accuracy of 80%. Since some error in the forecasted SKU demand should be expected, the accuracy can only go down from there.
Companies that have taken action to mitigate forecast error due to the financial plug have defined a maximum allowance for it. Trying to have no gap can easily become counterproductive, but a commonly used +/- 5% allowance is reasonable. This maximum allowed gap can drive action during the sales and operations planning (S&OP) process, when business leaders set the expectation that there must not be a gap outside the allowed window. When the demand planning team and key stakeholders have visibility to the revenue objectives and the latest monetized demand forecast, they know to start thinking of demand-shaping options as soon as they detect a gap that exceeds the limit. By the time the S&OP cycle comes to the executive S&OP meeting, demand planning stakeholders have a plan in place to make up the difference. The plan must include the products that they include in the demand-shaping activities.
It’s highly probable that the planning for the stretch demand at item details level will not be perfectly accurate. If the planning horizon is long enough — considering the supply lead times and time needed for the demand-shaping actions to bear fruit — the end item forecast accuracy is not critical at the far end of the planning horizon. The important thing is that there is a plan for specific products that at first can provide visibility to raw material and capacity requirements. As time goes on, the planning process will reveal opportunities to refine the plans and increase the accuracy so that by the time the finished goods accuracy is needed, the plans will be good enough. This way, there is no significant undefined plug hiding demand and revenue risk.
Marko Pukkila is VP, team manager in the Gartner Supply Chain Research group.