Open-Source MMM Is Shifting Measurement Value To Data

Circana announced on September 8 that it has added Google’s open-source Meridian framework to Liquid Mix, its self-service marketing mix modelling product. The company said the combination joins Meridian’s modelling approach with Circana transaction data, loyalty purchase signals, demographic information and retail outcomes.

The addition matters because it separates two parts of marketing measurement that vendors have often sold together: the statistical model and the commercial data needed to make it useful. As open-source frameworks become easier to adopt, CMOs may pay less for access to modelling code and more for verified inputs, calibration and the confidence to move budget.

What Circana Added

Liquid Mix itself is not new. Circana launched the platform in April and positions it as an always-on layer alongside traditional, consulting-led studies. The company says models can become operational in about a week, refresh daily and use point-of-sale data from more than 750,000 stores and 2,000 retail partners, supported by more than 10,000 benchmarks.

Meridian adds a transparent Bayesian framework that brands can inspect and operate. Google’s documentation says it can work with geographic or national data, incorporate reach and frequency, and use experiments to calibrate channel-level return estimates.

Circana chief executive Stuart Aitken said the integration gives clients “greater choice in how they measure performance” while connecting media investment to verified demand. The choice is meaningful, but the release also makes the division of value clearer. Meridian is available to anyone; Circana’s transaction and retail data are not.

New Forrester Wave Sets Benchmark for MMM Vendor Selection

Forrester's 2026 MMM Wave report benchmarks vendor selection across 31 criteria. Asian CMOs now have a practical framework to evaluate measurement partners as 53% adopt MMM as their core ROI tool.

Mission MediaMission Media

Where Measurement Providers Are Competing

Circana is not alone in making MMM faster and easier to operate. In July, Uptempo introduced a tiered product built on OptiMine that lets marketers start modelling with the data they already have. Jamie Glover, its head of strategic insights, said marketers need an approach that matches “where they actually are today.”

The competitive field is moving away from annual presentations toward software that teams can run between planning cycles. Providers are differentiating through proprietary data, benchmarks, workflow and services around increasingly accessible methods. That could give brands more leverage in procurement, but only if they distinguish the model from the surrounding vendor claims.

Why Open Source Does Not Remove Measurement Risk

Transparent code does not make every output causal. Meridian’s own guidance recommends calibrating return priors with incrementality experiments and warns that noisy data or weak assumptions can produce unrealistic estimates. Its channel calibration tools are designed to identify where uncertainty, bias or implausible return estimates justify further testing.

That distinction should shape the executive reading of Circana’s launch. A faster model can shorten the distance between evidence and a budget change. It can also accelerate a poor decision if promotions, distribution changes, pricing, stock availability or local market conditions are incomplete.

For regional marketing leaders, this is especially important. A common framework does not automatically make inputs comparable across Australia, Indonesia, Japan or India. Each market may have different retailer coverage, commerce platforms, promotional calendars and data maturity. Regional rollouts need a documented coverage map before one model output becomes the basis for cross-market allocation.

What CMOs Should Decide

CMOs evaluating Liquid Mix or another self-service MMM product should ask four questions before procurement: who owns the model, which inputs are independently auditable, how estimates are calibrated against experiments, and what confidence threshold must be met before money moves.

They should also keep decision rights explicit. Analytics teams can operate the model, agencies can interpret channel effects, and finance can test assumptions, but one executive owner must approve how evidence changes the plan. Otherwise faster measurement may simply produce faster disputes.

Circana’s Meridian integration suggests the market is moving toward more inspectable and accessible modelling. The practical advantage may no longer come from owning a proprietary equation. It may come from owning better evidence, knowing where it is weak and setting clear rules for when the result is strong enough to change the budget.