Do you also constantly monitor your forecast accuracy? Do you also ask yourself which key figure best measures forecast accuracy? Forget this way and do it right straight away!
In principle, there seems to be nothing wrong with measuring forecast accuracy, because if you don’t measure a target value, you can’t judge whether you have improved. However, looking at forecast accuracy often narrows your perspective and focuses all your actions on improving the quality of your forecasts instead of looking at the real target, which is delivery capability.
In fact, many companies still have considerable room for improvement in the quality of forecasts. Many do not use the mathematical possibilities available today and many try to improve the quality of their forecasts through “manual work”: the sales department is supposed to fix it and try harder or demand better data from the customers. This is where the thumbscrew of forecast accuracy seems to be just right.
Mathematics instead of manual work
However, it hardly makes sense to use the lever of forecast accuracy here, because you will not achieve systematically better forecasts with manual processes. There may be a small improvement under the thumb of forecast accuracy if everyone is pushed to be more diligent, but at what cost in terms of staff motivation and acceptance?
In order to systematically achieve better forecasts, you first have to start with the mathematics. Using statistics, simulation and artificial intelligence, special forecasting systems offer significant potential for improvement to companies that only use their ERP systems to determine forecasts. However, the forecasts will never be precise, because every market demand behaves chaotically to a certain degree and chaos cannot be forecast; it is therefore of no use to upgrade mathematically beyond a certain level.
Determining the accuracy of forecasts at this point doesn’t really make sense either, because if something can no longer be systematically improved from an economic point of view, there’s no need to keep measuring it; there’s no need to push algorithms to be more diligent.
Accept chaos as a factor
It can make sense to supplement “technical” forecasts with the assessments of sales. The question is what added value can be achieved by doing so. The comparison between the accuracy of the “technical” forecast and the accuracy of the “hybrid” forecast is the only place where measuring forecast accuracy might make sense.
It makes much more sense in all cases to measure the breadth of the chaos of market demand and to design safety stocks on that basis. Because: the battle for product availability can only be won through safety stocks, not through forecast accuracy.
