Market Mix Modeling: The Evolution and The Challenges

The key that opens the doors to a firm's sales is in the hands of the marketing team. Never has this been so true than now with the evolution of the digital ad channels in the early 2000's, and their subsequent maturity, coupled with the fact that the portion of the marketing budget allocated to "digital marketing" seem to be on an upside at the cost of the allocations to the more "traditional" channels. On one level this is quite exciting because the new landscape promises a better opportunity to measure the impact of the marketing dollars (as compared to the traditional analysis),  however, at another level, it presents two big challenges: (a) to monitor the health of the various ad channels, sometimes in real-time with daily deals and promotions becoming a new norm in certain industries, and (b) to come up with an "algorithm" to decide on the allocation of marketing budget to various channels. The first model that was employed by the digital marketers in order to solve these challenges was quite simplistic - the multi-touch attribution model with Last Click , First Click, Linear Attribution etc.  With time it became clear that effects such as cross-channel correlations cannot be explained using the multi-touch attribution analysis. Market Mix Modeling (MMM) was then employed to include cross-correlations, successfully. However MMM comes with its own set of challenges coupled with the ever increasing demands to provide a real-time analysis. Also, the ask for digital marketing promotions among different industries is quite different, and this leads to the use of disparate techniques under the broad umbrella of the market mix model. In this talk we will provide a review of the techniques being used in the various industries, and discuss the challenges faced by the market mix modelers.