Google expands capabilities of its open-source MMM program

Google updated Meridian, the open-source marketing mix model (MMM) it released in January. The program is designed to help marketers make better-informed budget decisions.

The updates expand what marketers can measure. Meridian now supports non-media variables, such as pricing and promotions, enabling businesses to have more information about how these factors impact sales. It also introduces channel-level contribution priors, which let marketers apply their own business knowledge to the model for more actionable insights.

In addition, Meridian now makes it easier to measure the longer-term effects of upper-funnel media — such as brand recall and delayed influence that drive purchases weeks later. This is powered by enhanced binomial adstock decay functions, which optimize over time. New marginal ROI (mROI) priors also help marketers identify where the next dollar will drive the highest return, based on past performance.

Dig deeper: Google releases its open source marketing mix model to all

Google first launched Meridian in the spring as an open-source MMM to help marketers allocate budgets more effectively in today’s complex, digital-first environment. After testing with hundreds of brands worldwide, the company made it widely available, emphasizing its transparency and customization.

Traditional MMMs struggle to accurately measure digital and AI-powered campaigns, particularly in channels like search. Meridian addresses these gaps with features such as Bayesian causal inference, direct integration with Google’s MMM Data Platform, and the ability to validate model outputs against real-world experiments.

Why it matters

Meridian offers a way to model performance across channels without relying on black-box vendor solutions. Being open-source, it gives data scientists transparency and the ability to adapt the framework. Google’s integration ensures access to more granular data for Google Ads, enabling more precise optimization.

Key features

• MMM Data Platform integration. Provides access to core MMM data for Google media, including search query volume for deeper insights.
• Customizable framework. Open-source structure allows marketers to adapt code and model parameters.
• Smarter budget allocation. Optimize based on sales, profit, or site visits using meaningful KPIs.
• Enhanced reach and frequency. Goes beyond raw impressions to account for the quality of exposure.
• Experiment-driven insights. Incorporates incrementality tests as priors for more accurate results.

Getting started

Marketers can download Meridian from GitHub and begin implementing it immediately. Google also works with more than 20 certified partners to support adoption, including firms like Analytics Edge.