Google Is Doubling Down on Meridian While Meta Quietly Shelves Robyn: What It Means for E-commerce Media Mix Modeling

Media mix modeling for ecommerce brands just split into two very different roadmaps.

Google keeps expanding Meridian, its open-source MMM tool, folding it directly into Google Analytics 360 and tying sales KPIs to adoption.

Meta, meanwhile, appears to be winding its own MMM tool, Robyn, down behind the scenes.

Sources across three agencies and three measurement vendors told AdExchanger that Meta’s Robyn engineering team has effectively been dismantled, even as Google keeps shipping new Meridian features every few months.

For teams currently evaluating or building workflows around either tool, that divergence matters this quarter, not next year.

Here’s what’s actually happening, what it signals, and why the tool you pick matters far less than the data feeding it.

What Robyn and Meridian Each Do, and Where They Differ

Both are open-source media mix modeling frameworks: statistical tools that estimate how much each marketing channel actually contributed to revenue, using aggregated data rather than user-level tracking.

Meta released Robyn first, back in 2020. Google was late to the category, launching Meridian in 2024, but has moved faster since.

The two tools reflect their parent company’s data ecosystem.

Robyn runs on R, a language more familiar to statisticians, and takes a more automated, budget-optimization-first approach.

Meridian runs on Python, uses Bayesian methods, and, as Google’s own data science lead has described it, maps out the “Googleverse” of media and data in intricate detail, particularly for YouTube and Search.

Why Meta Appears to Be Stepping Back From Its Own MMM Investment

According to AdExchanger’s July 2026 reporting, Meta has stopped actively pushing Robyn, and two separate sources used the word “dismantled” to describe what happened to the engineering team behind it.

Meta hasn’t abandoned measurement entirely.

In March 2026, it introduced a new “engage-through attribution” category and rebranded the metrics it ships to outside attribution platforms, a move one Meta product marketer said was meant to help the platform “tease out” the value of social media when advertisers lean on Google Analytics instead.

But the dedicated open-source MMM push looks to be over.

What Google Doubling Down on Meridian Signals About Its Measurement Roadmap

Google’s investment has accelerated all year.

In February 2026, Google launched a no-code Meridian Scenario Planner.

In May, it added Meridian GeoX for geo-level incrementality testing and Meridian Studio for managing large-scale models on Google Cloud.

Then, at Google Marketing Live in May 2026, Google announced Meridian would integrate directly into Google Analytics 360, combining first-party GA4 signals with external channel data from TikTok, Pinterest, and Snap inside one model.

Google’s sales teams reportedly have KPIs tied to Meridian adoption, and one measurement executive told AdExchanger the real driver is that Meridian upgrades Google Analytics, which has been trying to shake its historical dependence on click-based, last-click attribution.

That’s a reasonable business motive.

It also means Meridian’s rapid growth isn’t neutral. It’s Google building the on-ramp to its own ecosystem, in the open, which is a genuine improvement over a walled-garden dashboard, but not the same thing as a vendor-neutral measurement layer.

Why MMM Output Quality Depends on the First-Party Data Feeding It, Not the Vendor

This is the part that gets lost in the tool comparison.

Both Meridian and Robyn produce output that’s only as good as the inputs, and the platform default specs rarely fit a specific business without customization.

As one MMM vendor executive put it, Meridian “provides a good baseline but should be customized for every business,” and too many teams never do that work, effectively adopting whatever assumptions Google’s default configuration makes on their behalf.

The deeper issue is that neither tool was built to see beyond its own parent company’s walls.

A brand feeding Meridian clean, first-party, click-verified conversion data across every channel, not just Google’s, gets a fundamentally more trustworthy model than one relying on platform-reported inputs alone.

That’s true whether the model itself is Meridian, Robyn, or a proprietary MMM.

AdBeacon’s Meridian integration exists for exactly this reason: independent first-party data as the input layer, regardless of which platform’s MMM tool ultimately wins the adoption race.

What to Do If You Already Built Workflows Around Robyn

Don’t rip it out on the strength of one trend piece.

Robyn is still open source, still runs, and Meta hasn’t pulled the code.

What’s changed is the level of active investment and support behind it, which matters for how much you should expect future feature development, documentation, and platform-side data integration to improve.

If your models are stable and your team understands the assumptions baked in, that’s a reason to keep monitoring rather than migrate immediately.

What’s worth doing now regardless of which tool you’re on:

Audit which channels flow into your model and how. Neither Meridian nor Robyn automatically credits every channel accurately by default.

Feed the model first-party, verified conversion data wherever possible, rather than platform-reported numbers, since that’s the input quality gap that actually determines output quality.

Treat MMM as one leg of a triangulated measurement stack, alongside attribution for tactical signal and incrementality testing to validate both, rather than a single source of truth.

Revisit the decision in two to three quarters, once it’s clearer whether Robyn gets meaningfully less investment going forward or simply plateaus as a stable, lower-priority tool.

If you want to see what Meridian looks like powered by independent, click-verified first-party data instead of platform-reported inputs, book a live AdBeacon demo.

—-

FAQ

What is the difference between Google’s Meridian and Meta’s Robyn?

Both are open-source media mix modeling tools. Meridian, launched by Google in 2024, uses Bayesian methods in Python and is deeply integrated with Google’s own channels like YouTube and Search. Robyn, released by Meta in 2020, runs on R and takes a more automated, optimization-focused approach.

Is Meta discontinuing Robyn?

Meta hasn’t formally announced a discontinuation, but multiple agency and measurement vendor sources told AdExchanger in July 2026 that Meta has stopped actively pushing Robyn and that its internal engineering team has been scaled back.

Why is Google investing so heavily in Meridian right now?

Meridian is now integrated directly into Google Analytics 360, and Google’s sales teams reportedly have adoption tied to their KPIs. One measurement executive suggested this is partly because Meridian helps Google Analytics move past its historical reliance on last-click attribution.

Does the MMM tool I choose matter more than my data quality?

No. Both Meridian and Robyn produce output that’s only as reliable as the data feeding them. First-party, verified conversion data across every channel produces a more trustworthy model than platform-reported inputs, regardless of which MMM framework is running underneath.

Should I migrate off Robyn if I already built workflows around it?

Not necessarily right away. Robyn still functions and Meta hasn’t removed the code. Reduced investment affects future development and support more than current functionality, so auditing your data inputs matters more urgently than switching tools.