Measuring the true impact of your marketing spend is complex, especially when you look across channels and mediums. In Episode 6 of Google’s Ads DevCast, "The 'What' and 'Why' on Meridian," we sit down with Product Manager Katie Munro and Developer Relations Engineer Jeff Li to explore Meridian, Google's open-source Marketing Mix Modeling (MMM) framework, and how it helps advertisers make data-backed strategic budget decisions.
Watch the full episode
Hear the full episode
Here is a summary of the talking points from our discussion on what Meridian is, how the math works, and the newly announced surfaces to help you implement it.
What is Meridian?
Meridian is Google's open-source Marketing Mix Modeling (MMM) framework. It provides a holistic, privacy-safe view of marketing ROI by analyzing aggregated historical data across all channels (digital, TV, print) alongside non-advertising factors (seasonality, economic conditions). It helps you determine the incremental lift of your marketing (sales caused by ads) versus baseline sales (sales that would have happened anyway).
The Core Math: Hill Curves & Adstock
Meridian uses statistical models to capture the real-world behavior of advertising:
Hill Curves: Models diminishing returns, showing where you have room to increase spend (headroom) and where you have saturated a channel.
Adstock: Models the lagged effect and decay of advertising over time, capturing the delay between seeing an ad and making a purchase.
The Measurement Trifecta
Meridian completes the modern measurement framework:
Attribution (GA4): For day-to-day tactical campaign tweaks.
Incrementality Experiments (Meridian GeoX): To test specific hypotheses in the real world.
MMM (Meridian): For macro-level, strategic budget planning.
New Meridian Releases from GML 2026
We discuss three major additions designed to make MMM more accessible and accurate:
Meridian in GA360: A turnkey, integrated MMM solution within Google Analytics 360 that uses existing behavioral data to provide cross-channel ROI insights without requiring a data science background.
Meridian Studio: A Google Cloud-based enterprise platform to build and scale models, featuring native agentic AI to automate data quality checks, suggest model health improvements, and assist in scenario planning.
Meridian GeoX: An open-source geo-based incrementality solution that acts as a calibration engine, allowing you to run geographic experiments and feed "ground truth" causation data back into your Meridian model.
Next Steps:
