Executive Summary

Google previewed Meridian GeoX on May 5, 2026, an open-source, publisher-agnostic geo experimentation tool that feeds Bayesian priors into Meridian MMM. Testing begins later in 2026 and Google hasn't said when it will be available to everyone, so most brands can't run it in production yet.

GeoX is not the same as Google's Conversion Lift. Conversion Lift is a managed Google product that measures Google Ads inside Google. GeoX is open code that measures any channel against backend revenue you control. The methodology isn't even new, Google's trimmed_match and matched_markets repos have been on GitHub for years. What's new is the packaging, the name, and the explicit tie to Meridian.

The practical implication is straightforward. Open source moves the cost from license to labor, so running GeoX in production still requires data pipelines, an analyst, pre-test design discipline, and contamination monitoring. If your measurement team is one performance marketer, a managed platform is cheaper than self-hosting.

Either way, don't wait. Across 225 geo tests on Stella's platform, the median iROAS was 2.31x with 88.4% reaching significance. Brands running tests today will plug into GeoX immediately when it ships. Brands waiting will still be learning the basics in 2027.

That matters less than the trade press is making it sound, and more than most operators realize. Here's the practical version.

‍

What is Meridian GeoX?

GeoX is Google's open-source tool for running geographic incrementality experiments. It supports three designs: holdback, go dark, and heavy up. Results convert directly into Bayesian priors that calibrate Meridian MMM. GeoX is publisher-agnostic, meaning you can measure Meta, TikTok, podcasts, or any channel by geography, not just Google Ads. Testing begins later in 2026.

The mechanic is the same as any geo holdout. Split geographies into a treatment group and a control group, change media exposure in one, and measure the gap. What GeoX adds is structure around three specific designs:

• Holdback: keep ads active in most regions, pause them in a small set of holdout regions
• Go dark: pause media entirely in selected regions and measure the drop
• Heavy up: increase spend in selected regions and measure the gain

GeoX also runs multiple treatments against a shared control in the same study, which makes multi-cell tests cheaper than running them one at a time. The output feeds into Meridian as priors, so the MMM learns from real causal evidence instead of inferring impact from historical correlation.

A useful frame: Conversion Lift is the platform handing you a lift number. GeoX is the methodology, open and inspectable, that lets anyone produce one.

How is GeoX different from Conversion Lift based on geography?

Conversion Lift is Google's managed test, run through your rep, scored on Google's conversions, and limited to Google Ads. GeoX is open source, publisher-agnostic, supports any backend revenue source, and runs on geographies you pick. Conversion Lift answers "did Google Ads work inside Google." GeoX answers "did the campaign work in your business." They answer different questions, which is why they give different answers.

The two tools look similar from a distance and serve different purposes up close.

Google's Conversion Lift based on geography is a managed product. You request it through a Google rep. Google picks the geographic units (Google Marketing Areas), runs the test, scores it on Google's conversion data, and returns a lift estimate. It's a fast, free way to get a Google-blessed number for Google campaigns.

It also has the structural awkwardness of Google grading Google. The platform selling the ads also designs and scores the experiment that evaluates them. You wouldn't accept that arrangement from any other publisher.

Meridian GeoX flips the structure. The code is open and auditable. The geographies are yours to pick. The conversion data can come from Shopify, your CRM, Amazon, retail, or offline sources. The same framework measures Meta, TikTok, YouTube, podcasts, and CTV. That is what "publisher-agnostic" actually means: one methodology, every channel.

Quick comparison:

Should you wait for GeoX or run a geo holdout now?

Run one now. GeoX is in testing, and Google hasn't said when it will open up to everyone. Brands that already test will get value from it the moment it ships. The brands that don't test today will still be learning the basics when GeoX is finally available.

GeoX doesn't change the calendar problem. You still need spend data, revenue data, and geography aligned. You still need to pick comparable markets, control for contamination, and run the test long enough to detect a real effect. None of that gets easier by waiting for a Google product launch.

Across 225 geo-based tests on Stella's platform between August 2024 and December 2025, the median iROAS was 2.31x, the interquartile range was 1.36x to 3.24x, and 88.4% of tests reached statistical significance. The brands generating those benchmarks have already learned which markets behave reliably, which campaigns deserve test priority, and which contamination issues kill results.

Brands waiting until GeoX opens up to everyone will start that learning curve from zero. Sometime in 2027.

A useful sequence for the rest of 2026:

1. Pick one campaign worth testing. Branded search and Performance Max are the usual starting points because the gap between platform ROAS and iROAS tends to be widest there.
2. Run a four-week geo holdout. Use the Google Ads incrementality guide for setup, and the Performance Max-specific guide if PMax is in scope.
3. Document the result and the design. The result calibrates your budget. The design calibrates your next test.
4. When GeoX is available to everyone, port your geography logic and treatment design straight into the Meridian workflow.

The brands that win at measurement don't have better tools. They have a measurement habit. GeoX accelerates the habit. It doesn't create one.

‍

How does GeoX feed into Meridian MMM?

GeoX outputs become Bayesian priors that constrain Meridian's posterior estimates. In plain English, the MMM stops guessing from historical correlation and starts learning from your real experiment. The accuracy gain is meaningful. Stella's MMM runs 87% accurate on average in forward tests, and 95% when calibrated with iROAS from real experiments.

Most MMMs are pattern recognition. They look at historical spend, revenue, seasonality, and macro factors, then estimate which channels did what. That works until two channels move together, at which point the model has to guess. Guesses are usually wrong in expensive ways.

A calibrated MMM doesn't guess. It uses the result of a real experiment as a starting belief about a channel's true incremental impact, then refines that belief as new data arrives. That is what "prior" means here. The model isn't inferring from correlation alone, it's anchored to causal evidence.

The eight-point accuracy gain matters because budget decisions live downstream of the model. An MMM that's 87% accurate misallocates a meaningful share of every quarter's spend. An MMM at 95% misallocates less.

This is the part GeoX changes most. By making geo experiments cheaper and easier to run, more brands will produce the calibration data their MMMs need. For more on Meridian as an MMM framework, see Getting Started with MMM Using Google Meridian.

Worth saying plainly: Meridian is only better than the next MMM if you actually feed it experiment data. Most teams don't. GeoX makes that gap embarrassing instead of invisible.

‍

What does GeoX actually cost in practice?

Free in license, expensive in labor. GeoX is open source, so the code costs nothing. The real cost is the data infrastructure, the analyst time, and the operational discipline to run experiments correctly and translate them into budget decisions. If your measurement team is one performance marketer, a managed platform is cheaper than self-hosting.

Open source moves the cost from license to labor. That is true of every open source product, GeoX included.

To run GeoX in production, a brand needs:

• Clean data pipelines that join Google Ads spend (city-level), Shopify revenue (ZIP-level), and any other channels into a single geographic dataset
• An analyst who can configure the model, interpret Bayesian posteriors, and translate the result into a recommendation a CFO can act on
• Pre-test design discipline, because region selection is the single strongest predictor of whether a test reaches significance
• Contamination monitoring, especially around Google Ads location targeting defaults, which leak ads into "control" regions if you don't change the settings
• Maintenance, because the codebase will evolve and your version needs to keep up with the Meridian project on GitHub

If a brand has a measurement team, GeoX is great. If "measurement team" means one performance marketer with a Looker license, the math goes the other way. Stella's Incrementality product handles the design, contamination checks, and synthetic control matching, and the MMM runs at $3,000 per month flat instead of the $15K-$80K consulting fees attached to most bespoke setups.

The honest framing: GeoX makes self-hosting cheaper. It doesn't make self-hosting easy.

‍

Frequently asked questions

When will Meridian GeoX be available to everyone?

Google announced GeoX on May 5, 2026 and confirmed testing begins later in 2026. Google hasn't published a launch date. You can sign up for launch updates on the official Meridian GeoX page.

Is Meridian GeoX free to use?

Yes, the code is open source under Google's Meridian project on GitHub. The license costs nothing. The engineering, analyst time, and data infrastructure required to run it in production is where the real cost lives, which is true of any open source measurement tool.

Can GeoX measure channels other than Google Ads?

Yes. Google describes GeoX as publisher-agnostic. You can test Meta, TikTok, YouTube, podcasts, CTV, or offline media, as long as you can change media exposure by geography and pull a clean outcome signal. The same framework runs across every channel.

How long should a Google Ads geo holdout run?

Two to four weeks for high-volume DTC brands. Four to six weeks for mid-market. Add a two-week post-treatment window for YouTube, Demand Gen, or any upper-funnel test with delayed conversions. Run length should be driven by statistical power, not calendar habit.

Does GeoX replace MMM?

No. GeoX is the geographic experimentation layer. Meridian is the MMM. GeoX feeds causal evidence into the MMM as priors. The two work together: experiments validate causality, MMM allocates budget across channels and time. You need both for honest measurement.

‍

The bottom line

GeoX is the most important measurement announcement Google has made in years, and the most overrated.

It formalizes a methodology that good measurement vendors have been running for years, makes the open source code easier to use, and gives CFOs a Google-branded reason to trust geographic experimentation.

It doesn't pick your markets, manage your contamination, join your backend revenue, or interpret your results. Those are still the hard parts.

The brands that get value from GeoX when it ships are the brands running geo holdouts today. Everyone else will be six to twelve months behind on benchmarks, market matching, and the basic discipline of treating measurement as a quarterly habit instead of a one-off project.

Run your first Google Ads geo holdout before GeoX ships. Start a 7-day Stella trial and we'll match your markets, run the test, and have results ready before your next budget review.