Steve,
I would say that there are few advantages of true bottom up forecasting. In most demand planning designs the bottom is determined by the lowest level of the characteristic value combinations. (CVC's). in many cases this is very low, including levels like shipping locations and customers. When you are at a level this low you have two problems 1) the shear number of items to forecast is too many for any planner or group of planners to actualy look at let alone correct.) 2) the history at this level tends to be so sporadic that good statistical models are difficult.
I think the more correct approach is "middle up planning". In this case you pick a level above your lowest level that the two issues mentioned above are no longer a problem. In other words pick a planning level that the historical data is statistically stable and representative and the number of items to be planned can be managed by your planners. For the levels below that you need to rely on your desegregation factors to create a reasonable distribution of the forecast. You can also have exceptions to your rule where you might want to plan specific items at lower levels- your highest volume products at your biggest customers.
The trick in all of this is finding the right "middle level" and like goldilocks and the three bears you want a level that is not too high, not too low, but one that is "just right"
good luck finding it
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.