Adstock Transformations

The Adstock session in the MMM Advanced Workshop covers how advertising impact extends over time and reduces in effectiveness with repeated exposure. Adstock transformations are a key part of Marketing Mix Modeling, helping to realistically reflect how media influences consumer behavior beyond the initial interaction.

Key concepts covered include:

1. Carryover Effect

   • Media exposures don’t just affect sales instantly—they continue to influence behavior in following days or weeks.

   • Adstock captures this using decay functions to model the lingering impact of past advertising.

2. Diminishing Returns

   • With more frequent exposure, ads become less effective.

   • This concept ensures MMM models avoid over-attributing results to higher spend levels.

3. Types of Adstock Functions

   • Geometric Adstock: A simple exponential decay model, commonly used in MMM.

   • Weibull (CDF & PDF): Adds flexibility to model peak timing and slower or faster decay patterns.

   • Hill & Power Functions: Useful for modeling saturation and non-linear response behavior.

4. Business Relevance

   • Proper use of adstock improves ROI estimation and reveals how long different media types influence sales.

   • It enables marketers to better time campaigns and allocate spend more effectively.

5. Hands-On Learning

   • Participants compare how different adstock functions affect media contribution.

   • Visuals and examples illustrate how transformations change the interpretation of media impact in regression models.

By the end of the session, learners will be able to apply the right adstock transformation, interpret its effect on media ROI, and enhance model accuracy for better marketing decision-making.