Do the pieces fit? Assessing the configuration effects of promotion attributes

Despite the extensive resources allocated to sales promotions, managers are still often unsure which combinations of promotion attributes, in what circumstances, will be most effective in achieving their marketing objectives. This problem has intensified as more firms move online and engage in frequent promotions. We conduct two empirical studies using field data from hundreds of online campaigns to identify the significant interactions that lead to an immediate sales effect. We adopt a boosted tree (BT) approach to investigate how promotional attributes can be aligned with one another and with other contextual variables to achieve synergy, and propose a parametric test to assess the statistical significance of these interactions. Our findings provide valuable insights into effective promotional designs. We also compare the proposed approach with various machine learning methods to demonstrate the merit of the BT approach and its potential applications for strategic configurations in business and marketing.