Ebook
Understanding the CTV fragmentation challenge
Learn more
Loading Insights...
Many marketers struggle with cross-media measurement, especially in light of changes around third-party cookies. As digital channels grow increasingly more complicated in incrementality measurement, brands are searching for identity-based attribution solutions like data clean rooms.
Over the past decade, marketing measurement has changed dramatically. Not only do consumers have more devices than ever (according to Epsilon research, consumers now have 6.9 devices on average), but changes in the digital structure of the internet make it harder to anchor true measurement and attribution.
Third-party cookies, once widely used as a measurement indicator on the open web, have virtually gone away. Several browsers, including Safari, Internet Explorer and Firefox, deprecated third-party cookies years ago. Chrome remains the only hold out. In 2025, Google reversed its decision to phase out third-party cookies after struggling to find a strong identification alternative. Currently, Chrome holds more than 67% of the global browser market.
Without third-party cookies, marketers feared that digital advertising would be adversely affected by these changes. In a 2020 study from Epsilon that came out shortly after the announcement of Google's phase-out plan, 67% of marketers said they had negative feelings about the changes.
At the same time, the rise of walled garden data clean rooms run by Amazon, Meta and Google, means the data needed for measurement now lives behind privacy-preserving query interfaces controlled by the publishers, not in a brand's own analysis stack. These three tech giants make up over half of all revenue generated from digital advertising.
Data clean rooms powered by identity resolution promise to help overcome these key issues and avoid wasted spend. They can enable multi-channel, multi-touch attribution at scale across channels and platforms, including across platforms like Amazon, Meta and Google. These attribution models show the true impact of each channel and identify inefficiencies.
A data clean room is a safe, pseudonymized space for known and prospective customer data. This allows marketers to analyze marketing and advertising data from many different sources in one, singular view while protecting the data privacy of each individual source.
This is most helpful in marketing and advertising contexts, where brands often have their own first-party data, data from partners and platforms (like Meta, Google, etc.) and permissioned or purchased data from third parties that they're trying to resolve across each data source. Data clean rooms allow brands to sync all of these data streams into one view of each person across these different contexts, increasing the value of the information they already own.
But the right clean room doesn't merely become a warehouse for this data. It utilizes multi-party data, an identity layer and output controls that enable deeper analysis and aggregated insights to create a closed-loop system.
There are four key measurement use cases inside a data clean room.
| Use case | Short description |
|---|---|
| Closed-loop attribution | Exposure data and conversion data joined to measure the path from ad to purchase. This is provided in conversion rate, ROAS and cost per acquisition, and replaces what third-party cookies used to enable on the open web. |
| Incrementality measurement | A test/control comparison inside the clean room to measure incremental lift that is attributable to a specific media channel or campaign. This is the most rigorous measurement use case and requires the clean room to support holdout group analysis. |
| Marketing mix modeling (MMM) data feeds | The clean room produces aggregated, privacy-safe data feeds that flow into the brand's marketing mix model (MMM). This approach solves the problem of "the publisher won't give us the granular data we need," because the clean room produces the aggregate inputs the model needs without exposing individual records. |
| Cross-publisher reach, frequency and overlap | This is deduplicating audiences across multiple publishers/networks to measure unduplicated incremental reach and manage cross-publisher frequency. Critical for CTV and retail media programs across multiple retailers' media networks. |
Closed-loop attribution inside a clean room happens when brands can match exposure data with publisher/retailer conversion data, sometimes provided by direct integrations with publishers.
Clean rooms offer a privacy-safe space to join hashed identifiers via an identity layer. This produces metrics like aggregated conversion rate, ROAS or attribution tables.
An example: A CPG brand running a CTV campaign on a streaming publisher wants to know how many viewers exposed to its ad subsequently purchased. The clean room joins exposure and purchase data; the brand sees a conversion rate and ROAS, the publisher sees nothing about the brand's customer file, and neither party sees the other's individuals.
Incrementality distinguishes the true causal impact of an ad, revealing conversions that would have happened anyway and the ones the campaign actually caused (i.e., the added impact on the consumer from the advertising or marketing they received). This matters because it solves for inflated performance metrics caused by traditional attribution, and it better shows the how marketing specifically contributed to business impact, effectively boosting measured marketing ROI.
Clean rooms offer brands the ability to compare a test group (exposed) with a control group (unexposed), then match it to conversion data in a privacy-safe space. This environment allows partners to share data essential to this process and prove past simple correlation to verify conversion because of exposure.
It is important to note: Not all clean room platforms have this capability. It's important for brands to ask potential clean room vendors how they provide incrementality measurement.
Marketing mix modeling (MMM) has reentered the marketing mainstream as cookies have receded. Modern MMMs are hungry for granular, multi-channel data, and clean rooms are increasingly the pipe that delivers data without violating partner privacy contracts.
When aggregated, time-segmented exposure and conversion data flow into a clean room and through the MMM input layer, brands can quantify the relative contribution to sales of the various offline and online media impact at the market level. And this is the highest-volume, lowest-touch use case once it's set up because clean room queries run on a schedule and feed the MMM continuously.
Clean rooms can also solve the problem of measuring cross-publisher reach and frequency, typically a fragmented metric that's challenging to measure across various vendors.
How does this happen? Imagine a campaign runs across 10+ publishers. Each one self-reports reach, none of them deduplicate against each other and the frequency caps don't coordinate across vendors. Clean rooms can join exposure data from multiple publishers against a common identity spine to produce true unduplicated reach and frequency.
This is particular important for CTV activation and retail media programs.
While walled garden data clean rooms (DCRs) offer measurement, they often don't offer the level of granularity and use cases outlined above.
Walled-garden DCRs (Amazon Marketing Cloud, Meta Advertiser Cloud, Google Ads Data Hub) deliver deep measurement inside their respective ecosystems but the outputs are constrained and don't easily port across platforms. Neutral clean rooms (e.g., Epsilon, InfoSum) operate across publishers/retailers and produce measurement that's portable, but require partnership infrastructure that walled-garden DCRs include by default.
Most enterprise programs use both—walled-garden DCRs for in-platform measurement, neutral clean rooms for cross-platform measurement.
Measurement quality is fundamentally bounded by identity match rate. A clean room with 60% match rate and a clean room with 90% match rate produce measurably different ROAS numbers from the same underlying data—and the difference can change the media-mix decision the brand makes.
This is the single most important thing measurement leaders should evaluate when choosing a clean room platform: They shouldn't be most focused on the query language or the UI, but the identity layer underneath.
Identity resolution is key for marketers looking to bolster their campaigns. It allows brands to know their customers more deeply because they can see beyond their limited view of them.
The secret to success is using a clean room with data and identity pre-loaded.
Clean rooms pre-equipped with these features give brands a richer view of customers and prospects. They take a brand's customer data and augment it with the data already built in, making existing customer profiles more comprehensive and identifying potential buyers based on those insights.
That identity layer also lays the foundation for better measurement. When fueled by strong data and identity, and activated through a single source of truth, brands can rest assured that their measurement is accurate and granular.
When implementing sequencing for a measurement leader standing up a clean room measurement program, consider these five steps:
Epsilon Clean Room allows you to make the most of your data to drive better campaigns. is built with pre-loaded data that enhances your first-party data with identity resolution, giving you the best view of consumers on a platform for marketers and data scientists, all built on a privacy-first framework. Our person-based campaign activation enabled person-based reporting, meaning you can see how your campaigns perform on a micro and macro level. And our closed-loop measurement enables you to use insights gleaned today for tomorrow's campaigns.
Measurement that joins multi-party data (typically a brand's exposure data with a publisher's or retailer's conversion data) inside a privacy-preserving environment to produce aggregated metrics—conversion rates, ROAS, incremental lift—without either party seeing the other's row-level records.
Traditional attribution depends on third-party cookies or platform-provided pixels—both of which have eroded. Clean room measurement uses privacy-safe identity joins to connect first-party data across parties, which works in cookieless and walled-garden environments. The trade-off: clean room measurement requires data partnerships, contracts, and identity infrastructure that traditional attribution didn't.
No—it powers it. Clean rooms produce the privacy-safe, multi-party data feeds that modern MMMs need to model channel performance accurately. The MMM is still the modeling engine; the clean room is the data pipe.
Yes—Amazon Marketing Cloud is a clean room. The same is true for Meta Advertiser Cloud and Google Ads Data Hub. These are walled-garden clean rooms that deliver measurement inside their respective ecosystems. They're typically used alongside neutral clean rooms that handle cross-platform measurement.
Accuracy depends primarily on identity match rate—the percentage of records the clean room can confidently join across parties. Higher match rates produce more reliable conversion rates, ROAS, and incrementality estimates. The platform's identity layer is therefore the most important architectural choice for a measurement program.
