What is Marketing Mix Modelling?
Marketing Mix Modelling (MMM) is a top-down statistical method that quantifies the impact of diverse marketing investments on sales and market share, utilising data from advertising spend, promotions, pricing, and sales. Originally rooted in the “Four P’s” of marketing developed in the 1960s, MMM has evolved from traditional econometrics techniques focusing on TV advertising and promotions to incorporate more diverse data sources, automated data pipelines, and sophisticated statistical techniques, including Bayesian modelling. This evolution signifies a modernised approach to optimising the marketing mix, identifying effective marketing initiatives, and making data-driven budget allocation decisions.
What data is required?
Key data inputs for MMM include historical weekly sales, detailed marketing expenditure / GRPs across channels (TV, digital, print, radio), timing of marketing campaigns, pricing and promotional activity, and external factors like economic indicators, seasonality, and competitor activities. Bayesian modelling allows us a) to incorporate sources of information beyond the data - sector benchmarks, results from experiments, results from other models, into the priors to improve accuracy and reduce model development time, b) to design nuanced models that capture real world complexity and c) to generate outputs that include the degrees of uncertainty, allowing marketers to make decisions concious of their risk profile.
How does it work?
The MMM process begins with comprehensive data collection and cleaning, followed by exploratory analysis to discern patterns and trends. We use Bayesian models to unpick the relationships between marketing spend and sales, controlling for confounding variables. We use the models to create simulation-backed scenario planning tools and optimisations, helping clients identify marketing plans that achieve their objectives.
Where have we used it?
Our application of MMM spans various industries, including FMCG, retail, automotive, publishing, and technology sectors. The integration of Bayesian modelling techniques and automated solutions has not only enhanced marketing spend efficiency and sales growth for our clients but also provided a deeper understanding of market dynamics and uncertainties, leading to more strategic and informed business decisions.