When your ROAS goes flat, do you spend more or cut back? UK brand Crew Clothing faced this exact question. Their 30-day ROAS looked stalled, but when they measured long-term conversion effects through Meridian, a +70% uplift was hiding underneath. The measurement was wrong — not the campaigns.
What actually happened?
In May 2026, Google announced three new measurement infrastructure tools at Google Marketing Live. This isn't just a product update — it's closer to a declaration that in the AI era, measurement infrastructure is the growth engine, not ad creative. Google literally said: "Measurement is your engine for growth in the AI era."
Here's why this matters now. Third-party cookie blocking and iOS privacy changes left marketers running campaigns with broken measurement. According to Google Tag Gateway data, just closing that measurement gap recovers an average of 14% of conversions.
Here are the three core tools announced.
- Enhanced Data Manager
A visual interface connecting BigQuery, Google Drive, HubSpot, and Shopify. The new API lets you pipe offline sales and store visit data directly into Google's measurement signals — no coding required. - Meridian GeoX
An open-source solution for measuring geographic incrementality. It answers "did this ad actually drive a purchase?" through region-level A/B-style validation. Integrates with Meridian MMM and begins testing in H2 2026. - Meridian Studio
A Google Cloud enterprise platform for managing marketing mix models at scale. Large advertisers and agencies can customize and run high-volume models with significantly less time and resource overhead.
And the biggest headline: Meridian is now natively integrated inside Google Analytics 360. What was once a separate tool is now accessible directly from the Analytics 360 dashboard. Industry analysts are calling it "the most significant structural change since Meridian's global launch in January 2025."
Why did measurement become the core marketer skill?
Until now, there were basically two ways to measure marketing performance. Run short-term ROAS to check channel efficiency, or pay for expensive outsourced marketing mix modeling. Both had serious limits.
| Old approach | Meridian + first-party data | |
|---|---|---|
| Measurement scope | 30-day click-based ROAS | Long-term conversions + offline effects |
| First-party data | CRM/revenue data siloed | BigQuery, Shopify, HubSpot connected directly |
| Forecasting | Historical analysis only | QFC predicts conversions 6 months out (Gemini-powered) |
| Validation | Gut feel or outsourced A/B | GeoX geo-level incrementality testing |
| Accessibility | MMM = expensive agency project, months | Open-source Meridian + native in Analytics 360 |
The key new signal is Qualified Future Conversions (QFC). It's Gemini-powered and forecasts conversions up to 6 months ahead. In the Crew Clothing case, short-term ROAS was flat — but QFC showed 70%+ long-term conversion performance. It stops you from cutting campaigns that are actually working, just because the 30-day signal hasn't moved yet.
CMSwire called this shift "measurement moving from a reporting function to a foundational operational layer for AI-driven marketing systems." Google Tag Gateway fits the same story. By routing measurement requests through your CDN, it converts third-party signals into first-party ones — resulting in a 9–18% average conversion measurement improvement.
Meridian is open source
The core MMM model, GeoX, and Scenario Planner are all free on github.com/google/meridian. Shopify, ASOS, and Vestiaire Collective are already using it. If you're not at enterprise scale, the open-source version is the fastest way in.
Where to start right now
- Set up Google Tag Gateway (fastest win)
If you have server-side Google Tag Manager and a CDN (Cloudflare, Fastly, AWS CloudFront), connect them. Config only — no code changes. Average 14% conversion measurement recovery. Jarrah Agency has the clearest step-by-step guide. - Connect first-party data in Data Manager
Google Ads > Data Manager — connect BigQuery, Shopify, HubSpot, etc. If you have offline sales data, connect that too. Once live, the effect of online ads on offline purchases starts showing up for the first time. - Run your first MMM with open-source Meridian
Install from github.com/google/meridian. Python-based, comes with example notebooks. If your data is structured properly, you can run your first model relatively quickly. Analytics 360 users can now run Meridian directly in the UI. - Enable QFC to track long-term performance
Turn on QFC in Google Ads conversion settings. Gemini then forecasts conversion likelihood up to 6 months out. Once you've seen it, 30-day ROAS as your primary signal will feel incomplete. - Apply for Meridian GeoX beta (H2 2026)
When GeoX launches, you'll be able to validate incrementality by geography. Apply for beta through Google's official partner network (Adswerve, Brainlabs, Epsilon, etc.).
MMM requires solid data preparation
Marketing mix modeling needs at least 2–3 years of consistent marketing spend data, channel-level impressions and clicks, and revenue data. Bad data in, bad model out. Check your data integrity before diving in.
.png)
