New York Times Advertising and a major global wealth manager produced measurable targeting gains without third-party cookies in Q4 2025, running the publisher's first-ever data clean room collaboration through Decentriq - and the results have now been made public.

The case study, published by Decentriq and shared by industry observers in June 2026, details a campaign that matched subscriber data from The New York Times against customer records from a leading international bank, all inside a data clean room (DCR) environment that prevented either organization from ever accessing the other's raw data. The numbers are precise: a 20% uplift in Quality Visit Rate, a 61% increase in international click-through rate (CTR) lift, and a 25% performance uplift in the United States. The campaign was not a test. Driven by those results, the bank has officially renewed the program.

The challenge: wealth management audiences are hard to reach

The financial institution the case study covers is not named. According to Decentriq, it is a leading global wealth manager, and the campaign targeted households with $1 million or more in investable assets in the United States and $100,000 or more internationally. That is a narrow demographic by any standard, and reaching it without third-party cookies presented two distinct problems.

First, precise audience definition. Behavioral proxies - browsing signals, in-market indicators, demographic overlays - rarely surface high-net-worth households with the specificity a wealth management advertiser requires. Second, the addressability gap. Without third-party cookies to link publisher inventory to advertiser CRM records, the two organizations had no direct mechanism for finding shared users and building lookalike prospects from them. Clean room technology offered a way to close that gap.

According to Decentriq, the bank had already pioneered clean room workflows for several years before this campaign. For New York Times Advertising (NYTA), the advertising arm of The New York Times Company, this marked the first time it used a data clean room to collaborate on and activate its first-party audiences. That distinction matters. NYTA is a publisher with 12.33 million total subscribers as of Q3 2025, and digital advertising revenues that surged 20.3% year-over-year. Even with that subscriber volume and revenue trajectory, the publisher had not previously entered a clean room collaboration of this kind.

How the Decentriq data clean room worked

The technical architecture followed four sequential stages, each designed so that no participant could observe the other's underlying data.

Responsible data matching came first. Both datasets - the bank's customer records and NYTA's subscriber data - were uploaded to Decentriq's secure environment. The platform operates on a confidential computing model, which means data is encrypted even during processing. Neither the bank nor NYTA could extract the other's identifiers.

The second stage was audience analysis. Decentriq's platform measured the overlap between the two datasets. The finding was notable: 1 in 4 of the bank's customers are also Times readers. That 25% overlap confirmed, in quantifiable terms, that the Times subscriber base contains a material concentration of the bank's actual wealth management clients. It also established a logical basis for using those overlapping users as a seed audience for lookalike modeling.

Lookalike modeling was the third stage. The team created six distinct lookalike audiences from the overlapping seed population, testing two audience sizes:

  • Small/precise (1%): The tightest group, designed to mirror the existing client base as closely as possible.
  • Broad/expanded (5%): A larger group optimized for higher reach at the cost of some precision.

The 1% and 5% figures refer to the proportion of the broader addressable population included in each segment. A 1% lookalike selects the users who most closely resemble the seed - this is the precision end of the spectrum. A 5% lookalike expands that pool, trading off similarity for scale.

Finally, activation. The finalized audience segments were exported to Google Ad Manager (GAM), NYTA's ad serving platform. This step allowed the publisher to serve ads directly to the high-value prospects without ever accessing the bank's private customer data. Once in GAM, the segments ran against established control groups to measure incremental performance.

The entire pipeline - from data ingestion to segment export - stayed within Decentriq's environment. According to Decentriq CEO and Co-founder Maximilian Groth, "By using our data clean rooms, two leaders in their respective fields have proven that you can achieve record-breaking precision while remaining a responsible data steward."

The results: international markets outperformed most

The campaign demonstrated that clean room-modeled audiences outperformed traditional targeting, particularly outside the United States.

Quality Visit Rate measures how deeply users engage with a site after clicking - time spent browsing, pages visited, depth of interaction. The DCR-reached audience produced a 20% uplift on this metric, indicating that the precision targeting brought users who were genuinely interested in the bank's wealth management services, not merely users who clicked and bounced.

The international figures are sharper. DCR audiences outperformed the control group by 61% on CTR (0.89% versus 0.55%). International digital advertising is structurally harder to personalize than domestic campaigns because identity matching across different country-level data environments is more fragmented. The size of the outperformance suggests the clean room matching created addressability that would not have existed through conventional means.

In the United States, the most precise audience - the 1% lookalike - delivered a 25.12% uplift over the control group. The specificity of that figure, reported to two decimal places, reflects what the case study describes as "the value of precision." The 5% lookalike still performed at or better than the control lines, which is an important secondary finding. Expanding the audience to five times the size did not meaningfully degrade performance. According to Decentriq, this validates the inherent quality of NYTA's first-party data even when the matching net is cast more broadly.

Joy Robins, Global Chief Advertising Officer at The New York Times, framed the outcome in direct terms. "This collaboration proves that premium publishers can offer brands a responsible, high-performance alternative to cookies," she said. "Through our data collaboration in Decentriq's data clean room, we were able to uncover a meaningful audience overlap, revealing that 25% of the bank's customers are Times readers. This underscores the unique value our audience brings to financial advertisers."

Why this is the first, and what it took

For a publisher the size of the Times, the absence of clean room collaborations until Q4 2025 is not an oversight - it reflects the structural demands of operating one. Clean rooms require both parties to have their data in a compatible format, to agree on governance rules, to define what the clean room can and cannot output, and to find an activation pathway for the resulting segments. The bank, which had already pioneered this workflow for several years, was a prepared counterparty. NYTA needed a platform that could handle the ingestion and matching without requiring raw data exposure on either side. Decentriq provided that infrastructure.

The FTC issued guidance in November 2024 warning that data clean rooms are not privacy-preserving by default. The regulatory body emphasized that what differentiates a DCR from a standard data transfer is the design and monitoring of constraints - the rules that limit what analysis can be run and what can be exported. Without those constraints being appropriately implemented, a clean room can add new avenues for data leaks rather than reduce them. The Decentriq architecture operates on confidential computing principles where data is encrypted throughout processing, not only in transit and at rest.

The GAM export mechanism used in the campaign is notable. Decentriq's platform produced audience segments that could flow into NYTA's existing ad server, which means the activation did not require the bank to have any technical presence inside NYTA's systems. The segments existed as targetable groups inside GAM - anonymized cohorts rather than named individuals.

Context: a maturing clean room market

The NYTA campaign sits inside a broader expansion of data clean room infrastructure. NIQ launched its global data clean room on Snowflake on October 2, 2025, enabling marketers to enrich first-party data and measure campaign outcomes across global markets. LiveRamp added NVIDIA GPU infrastructure to its clean rooms in April 2026, upgrading compute capacity to handle AI model training at up to 15 times the speed of CPU-based environments. And when Publicis acquired LiveRamp for $2.5 billion in May 2026, industry analysts observed that the universe of unaligned clean room vendors narrowed materially - with Decentriq explicitly named as one of the remaining neutral alternatives.

That context gives the Decentriq-NYTA-bank collaboration additional significance. Decentriq is a Zurich-based company positioned as a neutral infrastructure layer, not affiliated with any holding company, publisher group, or demand-side platform. As the larger clean room vendors move inside holding company structures, the neutrality argument becomes more pointed for publishers and advertisers that do not want their data flowing through infrastructure owned by a commercial competitor.

The IAB Tech Lab's PAIR protocol, which uses commutative encryption to allow secure matching of encrypted user data without revealing underlying personal information, was formally released in version 1.0 in January 2025. Andrew Knox, Product Manager for Privacy Technology at Decentriq, had previously described the PAIR interoperability as "a great step forward for the industry." The NYTA campaign does not explicitly reference PAIR, but the workflow it describes - encrypted matching inside a DCR followed by segment export to GAM - is structurally compatible with that kind of cryptographic matching framework.

What it means for financial and luxury advertisers

The specific audience targeted in this campaign - households with $1M+ in investable assets - represents one of the most commercially valuable and technically difficult segments to reach programmatically. Wealth management advertising has historically relied on contextual placement (financial news environments, premium print adjacencies) rather than behavioral precision, because behavioral data at the individual level for high-net-worth households is structurally sparse and imprecise.

Clean room matching changes that calculation. If a financial institution already holds verified customer records for its existing client base, and a premium publisher can demonstrate a 25% overlap with those records, the seed population for lookalike modeling is both larger and more accurate than anything built from third-party behavioral signals. The 1% lookalike in the NYTA campaign delivered a 25.12% uplift over control. The implication is that the seed was high quality enough that even the tightest audience modeling produced material performance gains over a non-DCR control.

For publishers specifically, the case study validates a commercial argument that has been difficult to quantify. Premium publishers hold first-party data on engaged, authenticated audiences. But demonstrating that a subscriber base correlates with a specific advertiser's existing customer profile required either sharing data - which raises privacy concerns - or running a campaign and inferring correlation from performance. The clean room approach produces that correlation insight directly, before a campaign runs, through the overlap analysis stage. According to Decentriq, the 1-in-4 overlap finding was available before the segments went live - it informed the decision to proceed and shaped how the segments were structured.

The campaign's renewal also matters. A single campaign could reflect favorable conditions, an unusually well-matched seed, or a period of elevated engagement. A renewed partnership represents the bank's judgment that the methodology is repeatable and that the economics justify continued investment.

Timeline

Summary

Who: New York Times Advertising (NYTA), the advertising arm of The New York Times Company, and a major international bank identified in the case study as a leading global wealth manager, with Decentriq - a Zurich-based data clean room provider - supplying the technology infrastructure.

What: NYTA's first-ever data clean room collaboration, producing lookalike audience segments that were activated through Google Ad Manager. The campaign targeted households with $1M+ in investable assets (U.S.) and $100k+ (international) during Q4 2025. Results included a 20% uplift in Quality Visit Rate, 61% international CTR lift, and 25.12% U.S. uplift from the 1% lookalike segment.

When: The campaign ran in Q4 2025. The case study was published by Decentriq in June 2026. The bank has renewed the program following the initial results.

Where: The collaboration operated inside Decentriq's data clean room infrastructure, with segments exported to Google Ad Manager for activation. The bank is described as a global institution with U.S. and international operations. Decentriq is headquartered at Josefstrasse 219, 8005 Zurich, Switzerland.

Why: Both parties needed a method to match audiences and build lookalike prospects without third-party cookies and without exposing raw customer data. The clean room model allowed the overlap analysis to happen inside an encrypted environment, producing the audience segments needed to run a high-precision wealth management campaign while each organization retained control of its own data.