The art of the deal: Machine learning based trade promotion evaluation

Abstract

Trade promotions are a complex marketing strategy to drive up sales, involving retailer and consumer dynamics. Furthermore, these events are time-sensitive, influenced by past promotions and both competitor initiatives and responses. In the Consumer Packaged Goods (CPG) sector, the proportion of price-promoted sales to regular-priced sales has increased to a very significant level. Given their relevance to the manufacturer's revenue, proper promotional planning is crucial. In this context, this work proposes a decision support system capable of evaluating a hypothetical trade promotion's success, based on historic data, to be used for the promotional planning process of two key product lines of a CPG manufacturer. At the core of this decision support system, a predictive model, based on machine learning algorithms, will leverage both time series data and predictor variables, in order to better predict future promotional performance. This work pulls from many different branches of knowledge namely, Marketing, Economics, Forecasting, Machine learning and Data mining, areas which are briefly introduced.