When new products are introduced for a fast mover consumer good company it is important to be able to estimate the initial sales (first 3-4 months sales). The challenge is that here is no sales history to base the forecast on. In this context you have to help create a method for new product sales forecasting. I'm not interested in the Bass curve or other standard methods for this purpose. I'm looking for more sophisticated approach based on statistical simulation. I have a background as a statistical teacher so do not apply for this job unless you have in depth knowledge for this task.
My point of departure is to make cluster analysis of products exhibiting sames sales patterns. The clusters will be the base for making simulations for distributions initial 3 months sales. The approach of new products should be to identify the characteristics of the new product and then assign it to the specific cluster.Based on the cluster we will be able to assign a distribution initial 3 months of sales.
I'm looking forward to hear if anyone has knowledge in this area