Google Meridian GeoX: geo experiments and MMM for SMEs

From the promise of MMM to geographic verification: the missing leap

Marketing Mix Modeling (MMM) has returned to the center of marketing managers' agendas in the last two years. The reason is simple: with the decline of third-party cookies and the growing opacity of last-click attribution systems, companies are looking for statistical models capable of estimating the real impact of each channel. However, traditional MMM suffers from a structural limitation: it produces aggregate estimates, which are difficult to validate empirically.

Meridian GeoX was created precisely to fill this gap. How reported by Search Engine Journal , Google has globally launched this extension of the Meridian open-source framework, integrating the ability to run geo experiments directly within the modeling process. In practice, the results of geographic experiments feed the model, rather than remaining separate tools.

Therefore, the measurement-optimization-budget reallocation cycle becomes much shorter. This is not a cosmetic update: it is a paradigm shift in the logic with which campaign effectiveness is validated.

How Meridian GeoX works: the operational structure

Meridian GeoX is structured on two levels that work in synergy. The first level is the geo experiment : test and control geographic areas are selected, a campaign is activated or suspended asymmetrically, and the difference in results between the two groups is measured. This is the classic method of incrementality testing.

The second level is integration with MMM. The data produced by the experiment are used as prior Bayesian in the statistical model. In other words, the geo experiment provides an empirical anchor that reduces the uncertainty of the overall model estimates. Therefore, the model does not just infer the impact of a channel: it verifies it on a real geographic subset.

According to the official Google Meridian documentation , the framework is open-source and based on Python. Furthermore, it supports integration with first-party data, making it compatible with data architectures already in place in structured companies.

Geo experiment and incrementality: three concrete scenarios for the Italian market

For an Italian company with multi-region distribution, the practical applications are immediate. Below are three representative scenarios.

• Retail with regional stores: A chain with stores in Lombardy, Veneto, and Tuscany can test the incremental effect of a Google Ads campaign on a single region, keeping the others as a control. The result feeds into MMM and informs the budget allocation for the next quarter.
• E-commerce with geographic seasonality: a brand that sells products with variable demand by area (tourism, seasonal clothing, agri-food) can calibrate its media investments with local incremental data, rather than relying on national averages that hide differences.
• B2B Services with coverage by province: a professional services company operating in local markets can finally measure whether its digital spending generates incremental leads in a specific territory, separating the campaign's effect from background noise.

In all three cases, the advantage isn't just analytical. It's operational: budget allocation decisions stop being based on intuition or industry benchmarks and are founded on evidence produced by the company itself.

The context: why this tool is arriving now

The global arrival of Meridian GeoX isn't random. The ad measurement market is under pressure from multiple directions. On one hand, the deprecation of third-party cookies has reduced the reliability of multi-touch attribution. On the other hand, the proliferation of channels — search, social, CTV, digital out-of-home — makes it increasingly difficult to isolate the contribution of each.

Furthermore, the growing presence of AI in reporting systems introduces new complexities. As we analyzed in the context of Search Console AI reports and their impact on SEO audits , automatically generated data requires critical reading. Meridian GeoX goes in the opposite direction: it gives the marketer back methodological control over measurement.

Google's choice to integrate MMM and incrementality testing into a single framework addresses a real need: using only one approach leaves blind spots in ROI estimation, while combining them allows for cross-referencing results and reducing uncertainty.

Trade-offs and limitations to consider before adopting it

Meridian GeoX is not a plug-and-play tool. Before considering its adoption, it's useful to consider some concrete trade-offs.

Data requirements: geo experiments require sufficient volume per geographic area. In small local markets or those with low campaign density, statistical significance is difficult to achieve. Therefore, micro-SMEs with budgets under 5,000 euros per month might not have the necessary critical mass.

Technical skills: Meridian is a Python framework. Its implementation requires individuals with data science skills or, alternatively, technical partners capable of managing the setup and interpretation of results. It is not a tool designed for those who only use graphical interfaces.

Response time: a well-structured geo experiment typically requires 4-8 weeks to produce reliable data. Consequently, it is not suitable for short-term tactical optimizations, but fits into strategic planning cycles.

Spillover effects: Italian geographic areas are not isolated. An active campaign in Lombardy can influence demand in Piedmont or Liguria. Meridian GeoX provides mechanisms to model this effect, but its correct configuration requires methodological attention.

The visibility issue: MMM does not replace SEO

A common mistake is thinking that advanced measurement tools like Meridian GeoX are only about paid channels. In reality, MMM can include organic variables — SEO traffic, branded search, editorial presence — as model inputs. This makes the quality of organic visibility a factor that directly influences the estimation of overall ROI.

In this regard, the dynamics we described when analyzing the auto-expansion of AI Overviews and its impact on organic CTR become relevant even for those who build MMM models. If organic traffic decreases due to AI Overviews, the model must be able to interpret this correctly, otherwise it attributes to paid campaigns an effect that is actually an organic loss compensated.

Similarly, the correct representation of the brand in AI responses influences branded demand, which is a key variable in any MMM. As we explored in our article on risk of conflicting brand information in AI Search , weak editorial governance can distort demand in a way that is difficult to separate from media effects.

Recommended decision: who should explore Meridian GeoX today

Based on the analysis conducted, we at SHM Studio identify three company profiles for which exploring Meridian GeoX is justified in the short term.

The first profile is that of the mid-market company with a media budget exceeding 15,000 euros per month and presence in at least three Italian regions. In this case, the geographical granularity of the model produces measurable value and the volume of data is sufficient for reliable experiments.

The second profile is that of the omnichannel retailer that wants to measure the effect of digital campaigns on in-store sales by area. In this context, Meridian GeoX integrates with physical sales data and produces incremental estimates that digital attribution systems cannot provide.

The third profile is that of the company that has already invested in a data infrastructure — structured CRM, organized first-party data, internal analytics team — and seeks a methodological framework to leverage these assets in media planning.

On the contrary, for SMEs with limited budgets or without internal analytics expertise, the most effective path remains to first consolidate the quality of basic data and the consistency of digital presence. In this sense, a strategy structured SEO and campaigns Well-configured Google Ads represent the foundation on which to build advanced measurement models.

The bigger picture: AI measurement and visibility as a unified system

Meridian GeoX fits into a rapidly evolving ecosystem. ROI measurement is no longer separable from understanding how search engines and AI systems distribute visibility. As emerges from the analysis of the impact of AI Answers on citations and visibility for SMEs , traditional traffic and conversion metrics are losing their descriptive power. Tools like Meridian GeoX try to meet this challenge from the media measurement side.

However, the complete answer requires an integrated approach. Organic visibility, brand consistency in AI responses, content quality, and measuring average incrementality are variables within the same system. Those who manage them in a coordinated way gain a competitive advantage that is difficult to replicate in the short term.

To delve deeper into the strategic implications of this scenario and assess how Meridian GeoX fits into your company's digital roadmap, you can explore the entire section dedicated to SEO and AI visibility curated by SHM Studio, or contact our team through the page contacts for a personalized evaluation. Furthermore, for those managing LinkedIn campaigns integrated with geo-targeting strategies, our service of LinkedIn campaigns can complete the media picture to be modeled.