Circana shipped a new measurement product, Liquid Mix, built on Google’s open-source Meridian marketing-mix-modeling framework, the analytics firm confirmed this week through trade outlet FutureCIO.

What Circana actually released

Liquid Mix is Circana’s branded deployment of Meridian, the MMM (marketing mix modeling) codebase Google published in 2024. The pitch is straightforward: a privacy-respecting, server-side measurement stack that does not depend on user-level identifiers, which have been eroding under Apple’s App Tracking Transparency rules, Chrome’s cookie deprecation, and EU consent regimes.

The product positions against a measurement gap that opened after iOS 14.5: agencies and brands lost access to deterministic event-level attribution from walled-garden ad platforms, and the incumbents responded by tightening their own dashboards. Liquid Mix sells against that lock-in by offering an outside view of cross-channel spend effectiveness.

What Meridian brings to the table

Google published Meridian as an open-source library that runs Bayesian MMM on top of TensorFlow and TensorFlow Probability. It is built to ingest aggregated daily spend and revenue data, run causal-style inference without needing conversion IDs, and output per-channel ROI coefficients with credible intervals.

For Circana’s enterprise customers, the appeal is the modeling engine plus the data plumbing: Liquid Mix layers Circana’s panel data, retail sales feeds, and syndicated media inputs on top of Meridian’s statistical core. Brands get the open-source methodology without having to staff a Python team to maintain it.

Who is buying this category

Marketing-mix modeling has migrated from CPG back into the budget conversation at large advertisers. CPG and retail spend analytics on Circana datasets remain the core buyer base. Tech and telecom have re-engaged as the cost of audience targeting has risen. Financial services brands, under pressure to justify media budgets against revenue, are also running pilots, though Circana has not disclosed named accounts in the launch window.

What it costs and how it ships

Circana has not posted public list pricing. MMM vendors in this bracket typically charge six- to seven-figure annual SaaS fees tied to media-spend-under-measurement, plus professional services for onboarding and model calibration. Expect the same scale.

What the integration claims, and what it does not

Liquid Mix is positioned as complementing, not replacing, multi-touch attribution or platform-native lift measurement. Buyers running it still need a separate experimental design program (geo-tests, audience holdouts) because MMM alone cannot isolate incrementality. The Meridian framework helps here: it ships with calibration hooks for geo-experiment data, which is where the marginal accuracy gain comes from.

What to ask in a vendor demo

Procurement teams evaluating Liquid Mix should pressure-test four things before signing:

• Source data scope: which Circana panels, retail feeds, and media-cost exports feed the model, and at what cadence.
• Refresh latency: whether the system produces weekly coefficient updates or only monthly.
• Experimentation stack: whether geo-test calibration is included or requires a third-party tool.
• Lock-in risk: whether the underlying Meridian models and trained parameters are exportable, given Meridian is open-source.

What is still unknown

The launch announcement, as reported by FutureCIO, is light on specifics. Circana has not disclosed a general-availability date, named design partners, named channel integrations (CTV, retail media networks), or pricing tiers. A request for those details sent to Circana’s press office had not returned a response at the time of writing.

For operators, the headline test is whether Circana’s packaged version of Meridian can deliver calibrated MMM outputs faster than an in-house rebuild. Google’s codebase cut the engineering barrier; the remaining value is data, refresh cadence, and services.