Marketing Mix Modeling (MMM) is fundamentally a causal problem: its goal is to determine the causal effect of your marketing investments on your business outcomes. To do this rigorously, Meridian is built on a foundation of causal inference and Bayesian statistics. This section provides an overview of these core principles, explaining the "why" behind Meridian's methodology and the key assumptions that make its insights possible.
Causal Inference Foundations
This section introduces the core concepts of causal inference and Bayesian modeling, explaining why these approaches are essential for an actionable MMM.
| Page | Description |
|---|---|
| Rationale for Causal Inference and Bayesian Modeling | This page explains why Meridian is built on a framework of causal inference. It clarifies that the goal of MMM is to understand the causal effect of marketing activity on business outcomes. |
| A Primer on Bayesian Inference | Get an introduction to the core ideas behind Bayesian statistics. This page explains concepts like "priors" (what you know beforehand) and "posteriors" (what you learn from your data), and how this approach allows Meridian to provide a full range of uncertainty in its estimates. |
| About MMM as a Causal Inference Methodology | This page discusses how MMM fits into the broader field of causal inference. It compares MMM, which uses observational data, to controlled experiments and outlines the key assumptions that are necessary to draw causal conclusions from your marketing data. |
Meridian's Causal Inference Assumptions
For Meridian to accurately estimate the causal impact of your marketing, it must make a few key assumptions. This section details what those assumptions are and how they are justified.
| Page | Description |
|---|---|
| Required Assumptions | This page outlines the fundamental assumptions, like "conditional exchangeability," that are required for a marketing mix model to produce valid causal estimates. It explains what these assumptions mean in practical terms for your analysis. |
| The Causal Graph | Learn about the "causal graph," which is a map of the cause-and-effect relationships in your marketing ecosystem. This page explains how Meridian relies on this assumed graph to ensure that the model is correctly structured to estimate the causal effect of marketing spend on business outcome. |
| Estimating Incremental Outcome Using Regression | This page details the statistical method Meridian uses to calculate the "incremental outcome.", which is the underpinning of ROI, mROI, and response curves. This is the core of the analysis, showing you how much of your KPI is caused by each marketing activity. |
| Extension to Models with Reach and Frequency Data | This page explains how the causal framework is adapted for channels with reach and frequency data. It shows how these richer inputs allow for a more nuanced understanding of your media's causal impact. |
References
| Page | Description |
|---|---|
| References | This page provides a list of the key academic papers and research that form the theoretical foundation for Meridian's approach to marketing mix modeling. |