Differential Evolution Framework for Budget Optimization in Marketing Models with Saturation and Adstock Effects

Abstract

Marketing has proven to be an important asset to every business, and the growth of digital platforms has made it even more crucial, which makes the use of tools to measure and guide advertising strategies almost mandatory. In this context, Marketing Mix Modeling (MMM) has been seen as a powerful tool for backing up this process. However, the sole use of MMM does not help with complex budget allocation decisions, and some of the most used solutions for this modeling lacks on effective or intuitive solutions for this kind of problem. That said, the subject of this article is the proposal of a budget-allocation method consisting of a two-stage differential evolution optimization algorithm that considers the carryover and saturation effects for media channels in the calculations. The model was evaluated on a counterfactual scenario, aiming at the improvement of the Return on Ad Spend over a real scenario of an open database and showed positive results with a significant hypothetical increase in the financial result of the given scenario, proving to be a viable tool for this use.