Data collaboration that keeps information at its source

Q: How does WPP's data fusion methodology work, and why is it more suitable than key-based joins for bringing client audience segments into a shared planning infrastructure?

WPP's approach projects behavioral variables from a donor data set onto a target data set using nearest-neighbor matching on shared demographic and behavioral variables, rather than requiring a common unique identifier. Traditional key-based joins degrade quickly in privacy-regulated markets, where match rates between first-party and third-party data often cover only a small share of the intended population. The projection method extends usefulness of panel data well beyond what identifier matching alone can support.

Q: What is WPP's Open Intelligence platform, and why does keeping data at its source matter for building a data partner network?

Open Intelligence keeps data within the environment its owner controls rather than moving it into WPP's systems, so brands and data partners set the terms of access themselves and can revoke them at any time. That control is what convinces organizations who would otherwise avoid data-sharing arrangements, whether for competitive, regulatory, or contractual reasons, to participate. The design functions as the commercial mechanism behind the network's growth.

Q: How does WPP's Matched Market Testing capability enable causal measurement of campaign impact rather than descriptive reporting alone?

Matched Market Testing identifies test and control regions based on historical behavioral similarity, then applies geo-time regression to estimate what would have happened without the campaign. That counterfactual estimate is what turns the result into a form of incrementality testing: evidence of lift that can be validated by independent third parties. The method depends on selecting regions before a campaign launches, since evidence weakens once targeting decisions have already been made.

Keeping data under the data owner's control, rather than centralizing it for extraction, expands the range of brands and partners willing to join a data collaboration network, including those blocked by competitive or regulatory constraints from traditional data-sharing arrangements.

Adam Bailey

SVP, Head of Data Science, EMEA

WPP

What you'll take away from this session

Fragmentation was organizational, not a tooling gap

Decades of independent agency operation produced duplicated infrastructure and siloed data across markets. Resolving it required consolidating governance and restructuring teams alongside the technology itself.

Privacy-safe collaboration expands the data network

Keeping data under the owner's control expanded the range of partners willing to share it, beyond what extraction-based data-sharing arrangements typically allow. That architecture is what makes a large-scale data network commercially viable.

Self-service access without losing control

A governed catalog lets data scientists see available data sets, review approved use cases, and request access directly, without managing individual credentials. Governance and audit trails stay automatic as part of project setup.

Data fusion beats key-based matching

Nearest-neighbor projection of behavioral variables brings audience segments into planning without a shared identifier, closing a gap key-based joins can't. Match rates between first-party and third-party data often cover only a small population share.

Causal measurement needs upfront design

The Matched Market Testing approach is valuable before a campaign launches, when test and control regions are chosen using pre-campaign behavioral similarity. Adding measurement afterward produces weaker evidence of impact.

WPP Media's data and analytics organization grew through decades of agency acquisitions, leaving each constituent agency with its own cloud accounts, governance model, and data science tooling. That fragmentation was structural, produced by how the organization was assembled, and closing it required consolidating infrastructure and organizational reporting lines together.

The resulting audience intelligence platform, called Open Intelligence, keeps client and partner data under the owner's control rather than moving it into WPP's systems. That approach has opened partnerships with brands that would not participate in traditional data-sharing arrangements. Data scientists request access through a governed self-service catalog connected to Domino via service accounts, replacing informal, relationship-dependent data hunting.

To bring a client's proprietary audience segments into planning without a shared identifier, WPP built a data fusion pipeline using nearest-neighbor projection of behavioral variables from a donor data set onto a target data set. The pipeline is now packaged as a reusable workflow other market teams can run for their own clients, a foundation WPP is extending toward agentic tools that automate geographic audience visualization.

FAQ

How does WPP's data fusion methodology work, and why is it more suitable than key-based joins for bringing client audience segments into a shared planning infrastructure?

What is WPP's Open Intelligence platform, and why does keeping data at its source matter for building a data partner network?

How does WPP's Matched Market Testing capability enable causal measurement of campaign impact rather than descriptive reporting alone?

Transform the work that matters most

See how Domino helps the world’s most regulated enterprises build, scale, and govern AI-powered applications.

Watch demo Talk to an expert