Causal market share models in marketing: Neither forecasting nor understanding?

The results obtained by Brodie and de Kluyver are likely to be specific to aggregate bimonthly data. Aggregation eliminates useful sources of variation, and causes researchers to specify models with unrealistic constraints. Brand management needs to know the effects of marketing programs, which are obtainable from an analysis of store-level weekly data. Flexible models need to be specified and tested for this purpose.