[2P4-GS-10] AI application: marketing / optimization

[2P4-GS-10-02] Challenges of Individual Data and the “Rediscovery” of Marketing Mix Modeling in Marketing Science

〇Ryoma Yasunaga¹, Yusuke Kumagae¹, Ryo Domoto¹ (1. Hakuhodo DY Holdings Inc.)

[[Online]]

Keywords:Marketing Mix Modeling, Market Response Analysis, Time Series Analysis, Media Mix Modeling, Bayesian Statistics

Marketing science is an effort to improve decision-making quality in marketing by using scientific methods. Marketing science includes the analysis of individual data and that of aggregated data. Recently, methods using individual data have been especially actively studied. However, the barriers to collecting and analyzing personal data are becoming higher due to the trend of privacy protection and technical regulations. Marketing Mix Modeling (MMM) is a market response analysis method used to understand the return on investment of advertising and to optimize the budget allocation. MMM is a historical research area proposed in the 1960s. However, market research agencies and advertising agencies in the industry have built and provided their own MMM. The detailed technical specifications have been the proprietary know-how of each company. However, industrial players have released several papers and open-source software in recent years, and a new trend is emerging. This paper considers this as a “rediscovery” of MMM and discusses its history, research trends, and challenges.