---
id: google-meridian
version: "1.7.1"
license: unclear
license_treatment: unclear
maintenance: active
---

# google-meridian — Google's open source mixed marketing model library, helps you understand your return on investment and direct your ad spend with confidence. 

License: unclear · Maintenance: active · Downloads: 96.2K/mo

## What it is and what it does

Meridian is Google's open-source framework for marketing mix modeling (MMM), a statistical technique that isolates the causal impact of marketing channels on sales and other KPIs using aggregated, privacy-safe data. It answers questions like 'How much revenue did each channel drive?' and 'What is my marketing ROI?', then uses those insights to optimize future budget allocation. The framework is built on Bayesian causal inference and uses a No U Turn Sampler (NUTS) for MCMC sampling, which is compute-intensive but well-suited to GPU acceleration.

Meridian handles both geo-level and national-level data, provides built-in visualizations, and supports calibration with experiments and prior information. It requires Python 3.10+, 15 runtime dependencies (including TensorFlow, scipy, pandas, and xarray), and is designed to run on GPUs for practical performance. The package is actively maintained, has no known vulnerabilities, and is documented with tutorials and a detailed API reference.

Use it for:

- Measure the incremental revenue impact of each marketing channel (search, display, social, TV, etc.) to understand which channels drive the most value.
- Calculate marketing ROI and compare efficiency across channels to inform budget reallocation decisions.
- Optimize future media spend by using Meridian's budget allocation tools to simulate different spending scenarios.
- Validate MMM results against controlled experiments to calibrate model assumptions and improve forecast accuracy.
- Analyze geo-level marketing performance to identify regional differences in channel effectiveness and tailor strategies.
- Support privacy-safe marketing analytics without relying on cookie or user-level data.

## Worth the install?

AI-flagged interpretation of the facts above — verify before relying.

Meridian is a Bayesian marketing mix modeling (MMM) framework that quantifies the impact of marketing channels on sales and KPIs, enabling budget optimization and ROI analysis using aggregated, privacy-safe data.

Yes, if you have marketing data and the infrastructure to run it. Meridian is actively maintained, has no security vulnerabilities, and solves a real business problem (channel attribution and budget optimization) with a principled Bayesian approach. The main barriers are the unclear license (verify before production use), heavy compute requirements (GPU strongly recommended), and a large dependency footprint (15 runtime packages including TensorFlow and tfp-nightly). Install in a fresh virtual environment and plan for GPU resources.

## Install

pip install google-meridian
uv add google-meridian
poetry add google-meridian

## Installing google-meridian

Before you install: Low friction installation via pip; active maintenance with a recent release (23 days old). Requires Python 3.10+ and 15 runtime dependencies including TensorFlow, tfp-nightly, and tf-keras, which are compute-heavy. The project recommends a GPU (tested on T4 with 16 GB RAM) for practical use, though CPU operation is possible.

License in practice: License status is unclear—no SPDX identifier or explicit license text is declared in the package metadata. Before adopting Meridian in a commercial or proprietary project, verify the actual license terms in the GitHub repository to understand usage and distribution rights.

Quickstart:

pip install google-meridian
# For GPU (Linux with CUDA):
# pip install google-meridian[and-cuda]

import meridian
# Load data and instantiate a model
model = meridian.MMM(data, ...)

Requires Python 3.10+. GPU strongly recommended for practical model training; CPU-only operation will be slow. TensorFlow and tfp-nightly are heavy dependencies.

Verify before relying:

- Whether the unclear license permits commercial use or redistribution—check the GitHub repository for actual license terms.
- Exact GPU memory and compute requirements for typical marketing datasets beyond the T4 16 GB test case.
- Whether tfp-nightly (a pre-release dependency) introduces stability risks in production workflows.

## Package facts

- License: not declared (unclear)
- Python support: supports_current
- Install friction: low
- Maintenance: active
- Downloads: 96.2K/month (top 15,000 on PyPI)
- Known vulnerabilities: none known

## Tags

marketing mix modeling, MMM framework, marketing ROI analysis, budget allocation optimization, Bayesian causal inference marketing, channel attribution modeling, marketing impact measurement, marketing-analytics, bayesian-inference, budget-optimization

[View on SkillFed](https://skillfed.io/packages/google-meridian) · [View on PyPI](https://pypi.org/project/google-meridian/)