Fusing Conjoint Analysis and Sales Data to Simulate Promotion Effects

Conjoint analysis is a frequent used technique in marketing research. A limitation of current conjoint models, however, is that they are unable to include a time effect. One speciffic application of conjoint analysis is price sensitivity analysis (PSA), which in essence tests for consequences on preferences share because of price changes. This appli- cation can also be used to investigate the effect of promotions. Thus, given a promotion, the preference share will be determined, whereby not taking into account what has oc- curred before and/or after the promotion. An issue that often arises after a promotion is the so-called post-promotion dip. This dip is the result of the fact that consumers will engage in stockpiling and/or purchase acceleration because of the promotion. This is a serious disadvantage. Being able to make predictions over time could lead to competitive advantage for marketing researchers. This thesis attempts to improve current conjoint simulations by combining multiple data sources by means of data fusion. In addition a whole new simulation approach will be developed. Sales data has been estimated using an ADL(2,2) model and estimates were used to predict promotion effects which can be used in the simulation approach. The results from the simulation approach were much more dynamic as compared to the traditional method of simulation, whereby clearly showing also the long term effects of a promotion. Recommendations for future research are provided.