Why Collaboration Infrastructure is Now a Marketing Priority
AI-driven marketing demands better data than fragmented stacks can deliver. Max Groth of Decentriq explains how data clean rooms close the gap—with outcome evidence from two published campaigns.
According to Gartner’s 2026 CMO Spend Survey, 70% of CMOs consider AI leadership a critical goal this year. Only 30% say they have the data foundations to scale it.
That gap is an infrastructure problem, and in my experience working with enterprise marketing teams across Europe and beyond, it’s where most AI marketing investment stalls. Data clean rooms are the piece of that infrastructure that most teams haven’t yet addressed.
Data clean rooms and AI: at a glance
Three points first:
- AI-driven marketing requires high-quality first-party data; fragmented martech stacks have the data but can’t connect it.
- A data clean room enables brands and publishers to analyse shared first-party datasets without exposing raw data to either party, combining privacy with practical audience intelligence.
- Campaigns built on clean room infrastructure consistently outperform third-party targeting: published case studies show 80% higher engagement rates and 58% higher conversion rates.
What is a data clean room?
A data clean room is a secure, privacy-controlled environment in which two or more organisations can analyse overlapping datasets without either party gaining access to the other’s raw data.
Results are aggregated; individual records are never exposed. The underlying data stays with its owner. Only the analytical outputs are shared.
In practice, this means a retailer and a media partner can run joint audience analysis, measure campaign overlap, or build lookalike models on combined first-party signals, without transferring a single customer record between systems.
The analysis happens inside the clean room environment.
Why AI makes data clean rooms a strategic necessity
The numbers confirm what I see regularly in practice. Gartner’s 2026 CMO Spend Survey finds that organisations advancing fastest on AI allocate 21.3% of marketing budgets to AI infrastructure, against a 15.3% average.
They’re also the ones building the data foundations that feed model training. The survey’s own warning is clear: the risk is that CMOs invest in AI tools faster than they build the foundations, processes, governance, and talent required to scale them.
Stack complexity is compounding.
Martechvibe’s State of Martech 2026 report tracks the pattern clearly: a landscape where adding more solutions hasn’t reduced complexity (and where the tools teams do have are increasingly underutilised as a result).
AI personalisation and predictive attribution both require high-quality, consented first-party data to function well. When that data is siloed across a dozen platforms, models train on noise. This is a direct consequence of stack fragmentation as opposed to a talent or technology gap.
IAB State of Data 2025 captures this directly: nearly two-thirds of respondents cite significant challenges across data quality, data protection, and fragmentation among disparate AI tools. These aren’t edge cases. They’re the majority condition.
Attribution compounds the problem further. When a customer journey crosses multiple publisher environments, a brand’s CRM, and a programmatic layer, no single platform has complete visibility.
Models that attempt attribution without that joined view systematically underweight certain touchpoints and overweight others. The result is a media budget allocated against a faulty signal, repeated across every campaign cycle.
Clean room infrastructure is the structural fix. It creates a controlled layer in which first-party data can be analysed across organisational boundaries, so AI models receive better inputs at the activation stage.
What do data clean rooms actually change?
The clearest evidence I have comes from two campaigns run under comparable conditions. IKEA Austria ran an awareness and consideration campaign with willhaben, Austria’s largest digital marketplace, activating first-party CRM data through a Decentriq clean room.
Against traditionally purchased segments, cost per visit fell by up to 30%, with ROAS nearly doubling for the highest-affinity audiences. The clean room surfaced segments IKEA hadn’t known to target. That changed how they allocated budget.
Samsung, working with Publicis Media & United Internet Media, ran a cross-publisher campaign activating first-party CRM data through a Decentriq clean room. Using daily automated data matching via netID (a privacy-friendly identifier), Samsung reached 3 million existing customers and over 1 million potential new customers through lookalike segments, all within a 16-day campaign window and without relying on third-party cookies.
The clean room made cross-publisher activation at that scale possible while remaining fully GDPR-compliant.
Neither result is unusual for clean room deployments where the data match quality is high. What changes is the calibre of the audience signal available to the model. Clean rooms don’t replace AI; they give it something worth working with.
A harder truth I always raise with procurement and legal teams: not all data clean rooms are created equal, and it always comes out during the InfoSec due diligence.
The most rigorous deployments run on confidential computing infrastructure, where data is processed in hardware-enforced isolation so that neither party, nor the platform itself, can access raw inputs. That’s the foundation your AI activation sits on.
Where the real AI readiness work starts
The finding that stays with me isn’t the 70% who call AI leadership a priority. It’s the gap between that number and the 30% who say they’re ready to scale. That gap doesn’t close by adding more tools to the stack.
Building that relationship between a publisher, a brand, and the reader on a first-party basis is critical. Data clean rooms are the infrastructure that makes it possible at scale, with security guarantees that withstand legal scrutiny.
Martechvibe’s coverage of the AI readiness gap shows just how wide it is. The CMOs closing it are consciously making a structural bet that the collaboration layer is the missing piece, and that building it now creates an advantage their competitors won’t be able to replicate.
The ones still treating data infrastructure as a future consideration will find their AI tools running on worse inputs than those of the brands that moved earlier.
ALSO READ: XM Coach’s Co-Founder David Hicks on Building Trust, Resilience, and Business Value
