Google’s Meridian and the Future of MMM: Transparency is Great, but Prediction is the Endgame

Updated on February 17, 2025
In this blog

Share

Listen to this article

The launch of Google’s open-source MMM model, Meridian, is another significant step in making marketing measurement more accessible. Alongside Meta’s Robyn and other open-source initiatives, Meridian continues the trend of democratizing MMM, driving down costs, and putting pressure on the traditional black-box consulting firms that have dominated the space for decades.

This aligns with Keen’s philosophy: Marketing Mix Modeling (MMM) should be affordable, accessible, and scalable for brands spending as little as $250K per year—not just an expensive tool for Fortune 500s. Transparency in models helps build credibility, and open-source frameworks like Meridian will push the traditional MMM players to reveal their methodologies instead of keeping them locked away in proprietary systems.

But while Meridian (and other open-source models) offer an advancement for measurement, they don’t provide the full solution that modern marketers actually need. Because measurement alone isn’t enough—prediction is the missing piece.

The Fatal Flaw of Open-Source MMM: The Fox is Guarding the Henhouse

The fundamental issue with Google’s Meridian, Meta’s Robyn, or any other platform-driven MMM is the source of the data. Google and Meta—while champions of open-source modeling—are also ad platforms that directly profit from ad spend. When an advertiser runs MMM through Google, they’re still operating within a data ecosystem controlled by Google. The same goes for Meta.

This is the same issue we’ve seen when agencies run MMM for their clients—how unbiased can an analysis really be when the entity measuring performance also profits from the results?

Yes, Meridian will bring more transparency than the old closed-off systems, but brands need neutral, third-party MMM platforms that don’t have a financial stake in ad spend allocation. That’s where Keen stands apart.

Keen wins when our clients make smarter investment decisions—not when they spend more money with a specific platform. Our entire methodology is built on long-term profitability optimization, not short-term spend justification.

Open-Source is “Free” … But It’s Not Actually Free

A major selling point of open-source MMM is that it’s free—which is true, in theory. But let’s break that down:

Running an open-source MMM model requires internal resources like:

  • A data scientist or analyst to implement the code
  • A BI or visualization expert to interpret and present the results
  • A marketing strategist to connect the MMM insights to actual media planning
  • A finance or business lead to translate findings into investment decisions

In other words, open-source MMM isn’t “free”—there’s a real cost in time, talent, and decision-making inefficiencies.

This is why Keen is built differently. We allow marketers to run a Bayesian MMM in under 10 seconds. Simulate a probability-based marketing plan in 10 seconds. Go from data ingestion to full modeling and planning in under 30 minutes.

Brands need speed, scalability, and repeatability—something a complex, people-heavy open-source model simply cannot provide at scale.

The True Evolution: From Measurement to Decision-Making

Even if you solve for execution, open-source MMM still stops at measurement—and that’s not enough.

If marketers only use an MMM for post-campaign analysis, they’re assuming that what happened in the past will happen again in the future. That’s a flawed assumption. The past provides informed priors, but true marketing success requires scenario planning and predictive modeling.

This is where Keen completes the puzzle—by making MMM a living, breathing decisioning system.

Meridian tells you what happened. Keen tells you what to do next.

The Keen Advantage: A Full Macro & Micro Decisioning System

1. We Bridge the Gap Between MMM and MTA/Incrementality
Most marketers today have:

  • A macro MMM model for big strategic decisions (e.g., annual budget allocation)
  • Micro models like MTA or incrementality for execution-level decisions

The challenge? These models don’t talk to each other. Keen solves this by allowing brands to:

  • Take outputs from Meridian, Robyn, or any MMM
  • Feed them into Keen’s Bayesian framework
  • Validate insights using additional priors from MTA, lift studies, and incrementality tests
  • Refine models with real-world forecast reconciliation

2. We Make MMM a Continuous Process, Not a One-Time Report
Instead of a “set it and forget it” MMM model, Keen:

  • Runs real-time simulations to test marketing decisions before executing them
  • Continuously reconciles forecasts vs. actuals
  • Optimizes flighting strategies across channels, accounting for external factors like seasonality, economic conditions, trade promotions, and distribution shifts

3. We Optimize for the Right Metrics: mROI and NPV
Too many marketers blindly trust platform ROAS, but that’s not a true business metric. Keen focuses on:

  • mROI (Marginal ROI)—the return on the next incremental dollar spent
  • NPV (Net Present Value)—the long-term, profit-driven impact of marketing investments

4. We Turn Insights Into Actionable, Predictive Plans
Meridian helps measure marketing impact—but what marketers need is the ability to translate measurement into action.

This is why Keen’s platform doesn’t just analyze data—it helps predict the optimal plan for the future.

With Keen’s optimization engine, brands can:

  • Take MMM priors (from any model, including Meridian)
  • Simulate weekly flighted plans
  • Forecast expected revenue outcomes with an average 4% margin of error
  • Optimize marketing investments for long-term business impact

Final Takeaway: Open-Source MMM is a Step Forward, But Prediction is the Future

Google’s Meridian is a great evolution in transparency—but measurement alone doesn’t drive business growth. The real value comes from taking those insights and turning them into a predictive, repeatable system that continuously optimizes marketing investment decisions.

Keen exists to do just that.

Gone are the days of MMM being a one-time, gut-feel-driven, black-box analysis. We now have the ability to triangulate multiple models, leverage Bayesian priors, and simulate forward-looking scenarios to drive real, incremental business growth.

Welcome to the future of MMM! Ready to move beyond measurement to predictive marketing decisions?  Request a demo and explore Keen today.

Ready to transform your marketing strategy?