AI-Driven Marketing Mix Modeling: A Framework for Effective Budget Optimization

This dissertation explores the viability of AI and machine learning (ML) algorithms to aid marketing budget optimization, specifically through the application of Facebook's open-source Marketing Mix Modeling (MMM) tool, i.e. Robyn. The study begins with an analysis of the evolution of marketing theory, with an emphasis on the breakthrough innovations introduced by the digital technologies. It continues by examining the shift in consumer behavior, focusing on how the once linear path to purchase as theorized by the marketing funnel is no longer able to frame nowadays consumption and is being replaced by more contemporary frameworks such as Google's Messy Middle. Since new consumer behavior is characterized by a certain degree of "messiness", that is, complexity, this is where AI can come to the rescue. Traditional marketing budgets, that often rely on marketers' past experience and subjectivity, are too inflexible to deal with nowadays complexity; for this reason, after a brief introduction to ML and the MMM market, the study delves into the mechanics of Robyn within the R programming environment. Through a case study involving the dataset of a fictitious company, named Alpha Inc., the research demonstrates how Robyn's algorithm is capable of identifying an optimal budget allocation and measure returns across different marketing channels. Moreover, the findings highlight how Robyn can be successfully integrated with conventional marketing budget templates. The dissertation ends with practical insights, emphasizing the need for data quality, technical expertise, and time investment to leverage Robyn effectively. If on one hand Robyn provides a powerful framework for optimizing advertising returns, for it to achieve the best result human expertise is requested and necessary.