Unlocking the Future of Marketing Mix Modelling

Marketing Mix Modelling or MMM, currently a hot topic for numerous reasons, is facing a critical challenge in the digital era. The proliferation of platforms has made it increasingly difficult to establish a brand’s voice and personality. Simultaneously, the exponential growth of social networks has empowered consumers and given rise to countless new niches and audiences. Consequently, building a meaningful brand-audience relationship and accurately measuring its impact have become formidable tasks. The days of enduring relationships between brands and consumers have given way to rapidly shifting scenarios. As a result, Marketing Mix Modelling, the skill of measuring advertising campaign effectiveness, requires a refresh.

Data quality

  With shorter campaigns targeting smaller segments, the availability of data has decreased. Traditional statistical models and standard machine learning algorithms rely on large datasets. To address this limitation, new MMM techniques such as Bayesian models, Graph models, and causally explainable Deep Learning models need to be adopted. Synthetic data, which offers equivalent output sets to the input data, can also help generate larger sample datasets suitable for machine learning algorithms.

Funnel effects

  In a multi-channel environment where brand personality is built across numerous channels, funnel effects play a significant role. As the number of channels increases, spanning online and offline platforms, Marketing Mix Modelling becomes more intricate, and the interplay between different channels becomes inevitable. Existing MMM models often struggle to comprehend this mix, leading to undervaluation of certain channel effects. Understanding and accounting for funnel effects become crucial when establishing a brand personality across multiple channels.

Selection bias and causality

  The relationship between advertising and demand is no longer straightforward due to the abundance of channels, media, and audiences. Selection bias arises when an input media variable correlates with one or more hidden demand factors. Failure to identify these factors may result in erroneous interpretation of media contributions. New MMM models need the capability to incorporate other variables to accurately assess the effectiveness of different campaigns.

Measuring creativity

  In the digital realm, consumers have become highly engaged listeners. If a brand fails to communicate with its customers, another brand will seize the opportunity. Brands must constantly develop new creative strategies while finding efficient ways to measure their impact. MMM models should not only gauge channel contributions but also evaluate the effectiveness of different creative approaches.

By addressing these challenges and embracing innovative techniques, the future of Marketing Mix Modelling can be shaped to meet the demands of the evolving marketing landscape.