Marketing Mix Modelling

The goal of marketing mix modelling is to understand which parts of a marketing budget are generating sales. Organisations spend money across television, digital, radio, and promotions, often without a clear view of which channels are working. A marketing mix model uses historical sales and spend data to estimate the contribution of each channel.

The main work is building a regression model that explains past sales using marketing inputs, seasonality, pricing, and external factors such as economic conditions. The model is fitted on historical data, typically covering one to three years, and is used to simulate what sales would have looked like under different spending patterns. The outputs feed directly into recommendations about how future budgets should be split across channels.

The main tools are Python and R, using statistical modelling libraries. Bayesian regression methods are increasingly common because they handle limited data better than classical regression and produce uncertainty estimates alongside point predictions. Results are usually presented in Excel or a reporting tool, since the main audience is marketing and finance teams rather than technical ones.

Marketing mix modelling projects appear in consumer goods, retail, and media companies. In the Netherlands, large retailers and FMCG companies are the main employers for this type of work. The project connects to Data Scientist and Market Researcher roles, and often runs alongside Marketing Analytics or Pricing Analytics work.