Skip to contents

Media Response Modeling

Media Response Modeling (MRM) is part of a broader set of techniques and models under the Media Mix Modeling (MMM) umbrella. MRM focuses on modeling, quantifying, and analyzing the non-linear saturation (diminishing returns) relationship between media spend and/or (impressions/clicks/GRPs) and business outcomes, such as sales or conversions. These relationships are modeled as non-linear response curves that capture the diminishing returns of media investments. Establishing response curves for a mix of media is critical for optimizing the media mix for better performance, as it allows marketers to understand the effectiveness of their media investments beyond observed levels.

mrmopt: Media Response Modeling in R

mrmopt is an R package designed to facilitate media response modeling using Bayesian methods. It provides tools for fitting media response models, analyzing response curves, and optimizing media mix for improved performance. The goal of the mrmopt package is to make it easier for marketers and analysts to apply media response modeling techniques to their data, enabling them to make informed decisions about their media investments. The focus of the package is to comprehensively model the uncertainly around the response curves and build optimizations from a broader set of scenarios, rather than just optimizing for a single point estimate of the response curve. Bayesian estimation powered by brms package enables simultaneously a robust and flexible framework for modeling media response while remaining within the bounds of a set the non-linear response models commonly used (Logistic, Weibull, Gompertz, etc.). While flexible, the package is designed to be user-friendly and accessible to marketers and analysts who may not have extensive experience with Bayesian modeling, providing smart automations, defaults, and tools to guide customization.

Installation

You can install the mrmopt package from GitHub using the devtools package. If you don’t have devtools installed, you can install it from CRAN:

install.packages("devtools")

Then, you can install the mrmopt package from GitHub:

devtools::install_github("bdshaff/mrmopt")

Features

  • Fit Bayesian media response models using brms.
  • Analyze response curves to understand media effectiveness.
  • Optimize media mix for improved performance.

Documentation and Tutorials

Visit the companion website for detailed documentation, tutorials, and examples on how to use the mrmopt package for media response modeling: https://bdshaff.github.io/mrmopt/