Google Rolls Meridian GeoX Out Globally to Connect Geo Tests with Marketing Mix Models

# What changed Google announced a suite of updates to Meridian, its open-source marketing mix modeling tool. The headline is the global launch of Meridian GeoX, which moves the geo-based incrementality testing tool out of beta and into general availability. Google also added model-building assistance labeled as agentic capabilities and expanded the kinds of brand signals Meridian can include.

# What GeoX does and why it matters GeoX is designed to run geographic experiments that estimate the incremental impact of advertising. Instead of relying on user-level tests, GeoX creates treatment and control regions to measure whether media spend caused real lifts in outcomes. It works across advertising platforms, so marketers can test campaigns on non-Google channels or combinations of channels.

The practical gain: Geo experiments produce causal evidence that can reduce uncertainty in MMM estimates. Marketing mix models use historical patterns to attribute outcomes to channels, and those patterns can be confounded when many channels or external factors shift at once. GeoX offers a ground-truth test marketers can feed into the model to increase confidence in specific investment effects.

# Meridian MMM Google says GeoX outputs can be incorporated into Meridian models as another source of evidence. That integration tightens the connection between controlled experiments and modeling: Meridian provides a broad historical view, while GeoX supplies localized causal results. This combination helps when teams must decide how much confidence to place in model-driven budget or strategy decisions.

# New model-building assistance (agentic capabilities) Meridian now includes tools that help during the modeling workflow. These capabilities can audit data quality, identify and help resolve errors, and offer guidance in real time while a model is being built. The goal is to reduce manual troubleshooting and speed up analyses by catching problems in the underlying data before they block an analysis.

Back-end changes were also made to run analyses faster and more efficiently. These improvements do not change the data inputs marketers must provide: teams still decide which business and media metrics belong in the model and how to interpret model confidence.

# Brand signals and longer-term effects Meridian can now accept relevant brand signals, including Branded Google Query Volume. These signals aim to surface upper-funnel changes that precede conversions, such as the delayed effect a TV or out-of-home campaign can have on demand. Branded query volume provides an observable proxy for branded interest that the model can use alongside media and sales data.

Caveat: an increase in branded searches does not by itself prove causation. Competitor actions, seasonality, promotions, and news coverage can all affect branded demand. Meridian treats branded volume as an additional signal to be interpreted in the context of the full model.

# What advertisers should consider GeoX makes geographic testing more accessible for teams that need causal evidence without user-level experiments. Using GeoX results to calibrate MMM can make model outputs more actionable for budgeting and cross-channel planning. The agentic tools reduce some technical friction, though teams remain responsible for selecting inputs and judging model confidence. Including branded query signals broadens the types of marketing questions Meridian can address, especially for campaigns with delayed impact.

# Bottom line Meridian's updates focus on linking experimental causal evidence to broader modeling, improving the model-building workflow, and bringing in signals that capture longer-term brand effects. For marketers who rely on MMM, the global availability of GeoX and the new tooling aim to reduce uncertainty and speed decision-making.