Meridian Open-Source
Build, run, and analyze Marketing Mix Models (MMM) with Meridian's advanced Bayesian modeling and causal inference capabilities.
New & Noteworthy
As of Sept 3rd, Meridian (v2.0) is now available with key updates to the library:
- Easily ground models in real-world proof with native geo-experimentation (Meridian GeoX) and automated calibration to bring causal lift results directly into your MMM. Plus, expanded model health diagnostics to understand the impact of calibration and see recommended channels for lift testing. Check out this demo for details.
- Capture the full effect of Brand building with our Full-Funnel MMM framework. Measure the indirect effects of upper-funnel brand investments on sales by incorporating Brand Equity signals like branded Google Query Volume (bGQV) directly into your models. See the Colab notebook for details.
- Accelerate modeling and get expert guidance with our agentic skills repository. This MCP-based toolkit empowers you to use AI agents to automate manual tasks, resolve errors, and get step-by-step guidance right in your terminal. With seamless access to Meridian expertise, you’ll be able to build higher-quality models, faster.
- Generate insights faster with Meridian’s new JAX backend and enhanced optimization engine. Meridian now runs 2x faster and 4x more efficiently and enables you to generate custom Scenario Planners in minutes. Empower your marketing teams with the tools to make better business decisions.
Explore the Meridian Open-Source Library
Basics
Introduction to Meridian, glossary, and FAQs.
User Guide
Step-by-step instructions for installation and using the Meridian library.
Modeling guides
Get guidance on every step of your Meridian journey:
- Pre-modeling: Gather the right data and prepare for modeling.
- Applied Modeling: Set up and run Meridian, including advanced customization and calibration.
- Post-modeling: Evaluate model health, interpret results, and optimize budgets.
- Bayesian Modeling & Causal Inference Theory: Explore the theoretical foundations of Meridian.
Code Examples
Explore end-to-end examples and use cases in interactive Colab notebooks.
API Reference
Detailed documentation for all classes and functions in the Meridian library.
Changelog
Stay up to date with the latest releases, new features, and bug fixes in the Meridian library.
Educational Video Series
-
Introduction to Meridian
-
Demo of Meridian
-
Geo vs National Level Modeling
-
Introduction to Priors
-
Treatment Prior Types
-
Calibrate Treatment Priors
-
Knots in Meridian
-
Incremental Outcome, ROI, mROI, Response Curves
-
Controls, Mediators & Treatments in Meridian
-
Adstock and Hill
Recommended Learning Paths
Meridian is designed for cross-functional measurement teams. Depending on your role, we recommend the following paths:
Marketing Analysts & Business Users
Start with Pre-modeling to collect and organize your data. Then explore the Post-modeling guides to interpret visualizations, evaluate ROI, and run budget optimizations.
Data Scientists & Technical Practitioners
Dive into Applied Modeling to configure the Bayesian model and customize priors. Explore Bayesian Modeling & Causal Inference Theory to understand the mathematical and theoretical foundations.