Case Study: Marketing Mix Modeling (MMM)
THE PROBLEM
Our client aimed to maximize new customer acquisitions by identifying the most cost-effective media channels and optimizing their media budget for the upcoming year.
OUR CHALLENGE
To achieve this goal, we needed to develop a robust, data-driven model capable of accurately attributing new acquisitions to various media channels.
This involved accounting for diverse variables like seasonality, economic factors, and both internal and external business drivers.
Additionally, the client required a solution that could adapt to evolving market dynamics and be regularly updated.
OUR SOLUTION
We built a custom, nonlinear probabilistic model that:
Assigned attribution to each media channel (TV, Radio, Facebook, Search) and controlled for variables such as seasonality, economic factors, and externalities.
Matched the client’s daily or weekly media spend to target KPIs using over two years of historical data.
Provided detailed insights into each channel’s efficiency, including diminishing returns and the adstock effects—how long the channel’s impact persisted over time.
Integrated submodels for demand-capture channels (like branded search) to enhance attribution accuracy.
THE RESULT
The model achieved impressive accuracy, with a predictive error rate of just 4.0%, and revealed key insights:
Facebook and TV: Both channels were highly cost-efficient, with Facebook directly driving the highest volume of new acquisitions, while TV contributed significantly to branded search volume and boosted brand awareness.
Radio: Although less cost-efficient, radio's longer adstock effect allowed for strategic ad hiatuses, lowering overall CPA.
Leveraging these insights, we built an optimization tool on top of the MMM results, enabling the client to allocate their media budget in a way that maximized new acquisitions within a set spending limit.
LOOKING FOR MORE RESULTS?
