Email
sinha.rk.dr@gmail.com
Why use sophisticated models when the input data is stale?
FMCG companies are moving to more and more sophisticated models of demand forecasting. They believe it will improve forecast accuracy and reduce the demand-supply mismatch. Unfortunately, they don’t see any improvement in their forecast accuracy.
The main reason for poor accuracy is not the models, but the forecasting process itself. This month’s S&OP would have used demand data till July for forecasting September, October and November.
If you forecast for September, why use data till July only? It’s not that August data is not available. We are just being blind to it, living with demand latency and blaming the models.
Let’s use the latest data for predicting demand. If it means doing it more often, so be it.