Google updates Meridian MMM for ROI planning - what changes for measurement and allocation

What’s new

• Support for non-media variables such as pricing and promotions. See Meridian.
• Channel-level contribution priors to encode expected KPI shares by channel. See channel-level contribution priors.
• New adstock_decay_spec parameter with binomial decay for longer-term effects. See binomial adstock decay functions.
• Geometric decay remains available for short-term carryover. See the same adstock configuration guide.
• Marginal ROI priors to guide allocation using incremental returns. See marginal ROI (mROI) based priors.
• Documentation contrasts mROI and ROI priors with selection trade-offs. See mROI vs ROI priors.
• Google named 30 new certified global partners for Meridian implementations.
• The site adds advanced modeling guides and examples. Start with Meridian.

Why it matters

These additions strengthen Meridian’s planning and calibration options, helping teams estimate true incremental impact and make smarter allocation decisions. In a market where marketers need a clear view of what drives ROI, more flexible priors and improved adstock modeling can reduce guesswork and speed up decision cycles.

How it works

Meridian is an open-source MMM that uses aggregated data and statistical modeling to estimate media and non-media contributions to KPIs. Priors let teams encode business knowledge and constraints, such as channel-level contribution priors or mROI-based priors for next-dollar optimization. Adstock functions model delayed effects of advertising, with both geometric and binomial adstock decay functions available. The documentation includes examples, parameter references, and guidance for calibration and model selection on the Meridian site.

Getting started

Review the latest guides and examples on Meridian, then consider engaging one of Google’s 30 new certified global partners for implementation and support.