Improving promotional effectiveness for consumer goods—A dynamic Bayesian approach

Abstract

AbstractA leading global personal care manufacturer was spending millions of dollars towards giving promotional offers on various goods throughout the year. The C‐suite (i.e., CEO, CFO, COO, and CIO) of the global personal care firm was aware of the increasing promotion spends year on year and this led to questions on promotion effectiveness. The main question faced by the firm was whether the returns they were getting in terms of revenue was justifying the increase in promotional spends. In addition, they had the following concerns: (i) Are they evaluating their promotion effectiveness in a scientific way? (ii) Given that the nature and intensity of promotions differ across weeks, regions, and the size and type of retail outlet, can they quantify promotion effectiveness for different time periods and at different levels of hierarchy of retail outlets? (iii) Can we develop a tool to evaluate and optimize promotional spends periodically? Based on our previous association with this firm, we were approached to solve their problems related to promotions. After our initial meeting, we realized that our solution should consist of four pillars: (i) Data—understanding the data and data wrangling, (ii) Model—developing a statistical modeling framework to evaluate promotion effectiveness, (iii) Computation—coding with emphasis on computational speed in order to update returns periodically, and (iv) Dashboard—producing an interactive tool for viewing results and carrying out budget optimization by stakeholders. In this article, we consider one brand of the manufacturer as an example to present our solution as a case study. This brand under focus is a face care brand in India and it spends about USD 4 million every year on promotions. We model the promotion effectiveness under the framework of dynamic linear models (DLM) using integrated nested Laplace approximation (INLA), a fast, approximate computational framework for Bayesian modeling and prediction. Companies like our personal care manufacturer belong to an industry which is commonly known as consumer packaged goods (CPG). An overview of this industry is presented in the following section.