Over the past two years, data clean rooms have exploded onto the programmatic advertising scene, and they’re already at the center of some of the most exciting new partnerships and growth opportunities.
But despite their rapid adoption, the definition of what a data clean room is – and all of the related nuance – is not well understood.
Retailers, broadcasters and walled garden platforms are among the first movers. In some cases they’re partnering with independent third-party clean rooms and in others they’re using clean rooms built within a walled garden. Some clean rooms enable only measurement and analytics, while others can be used to target ads.
At their most basic, data clean rooms are privacy-based technologies that allow companies to merge and match two or more first-party data sets in order to create a new audience or analytics segment informed by both data sets – but without ever allowing either side’s personally identifiable information to be exposed to the other.
It’s a compelling proposition for privacy-minded advertisers in light of ongoing platform changes and regulatory scrutiny. But please don’t call it a “DCR,” said David Danziger, who moved to clean room tech provider Habu last year after more than eight years as VP of partnerships at The Trade Desk. (We’ve got enough acronyms.)
“I’m desperate for DCR not to become a three-letter acronym for ‘data clean room,’” Danziger said. “But ‘data clean room’ seems to be the terminology that’s caught on for the general concept of first-party data collaboration among multiple parties.”
The initial platform play
The earliest data clean rooms and what industry execs typically think of as clean rooms are the data-sharing products created by the major walled gardens – namely Google’s Ads Data Hub (ADH), the Amazon marketing clean room and Facebook Advanced Analytics.
The major walled gardens seized on clean room tech as a way to continue targeting and attributing campaigns based on first-party data, but without exposing user-level data to vendors or advertisers.
The first major breakthrough entrant was Ads Data Hub. Since 2018, ADH has been the only way to use Google ad server data in Europe, because it would be a GDPR violation if advertisers could retarget Google users without explicit consent.
Google advertisers upload their first-party data to the Google cloud warehouse, called BigQuery. Google can then use its own massive audience graph to enhance the advertiser’s data with its own data from products such as Maps, Chrome, Gmail, YouTube, Search and the Google ad tech stack. This is possible because ADH doesn’t send user-level data back out to the advertiser.
The Google ad server tracks audiences (on Chrome, at least) all over the web with third-party cookies. Although user-level tracking is not possible through ADH, the data used in the campaign is richer because it incorporates Google’s own audience data.
Rather than having 10,000 user-level IDs that can be retargeted or tied to an individual, for example, an ADH campaign might inform an advertiser that, say, 5,000 of those new customers were added in midwestern cities like Chicago and St. Louis. Since Search and YouTube data is included in ADH, too, Google might inform a restaurant chain that its new customer wins come disproportionately from people who follow local sports avidly on YouTube or from people who recently searched for tickets to games.
Advertisers can’t add these insights to their CRM, but it’s useful analytics all the same.
In 2020, ADH took a big step from pure analytics to allowing audience activation, aka, targeted ad buys. But since the data is enriched with Google’s first-party IDs, the only DSP allowed to use ADH audiences is Google DV360.
Likewise, Amazon’s cloud-based clean room can only be used for analytics or within Amazon’s own DSP.
Independent clean rooms
But although walled garden clean rooms may be most prominent, they’re poor examples of how data clean rooms can and should work, according to Steve Silvers, Neustar’s SVP and GM of marketing solutions.
ADH, for instance, primarily serves Google. Advertisers load their first-party data into Google Cloud and Google uses it to improve targeting and attribution. But this process doesn’t truly enhance the advertiser’s data.
An independent data clean room is operated by a standalone third party with no stake in the ad campaign. Its only purpose is to warehouse data and perform a match across different data sets.
The incumbent in the category is Snowflake, with LiveRamp’s Safe Haven, TransUnion-owned Neustar, Habu and startups, including InfoSum, in the mix as well.
Using an independent clean room, advertisers can load their data to the cloud along with any supply-side or other data partners and sync their audiences. The advertiser is able to use an open web DSP and both the advertiser and sell-side data partner in the clean room are able to enhance their first-party data sets.
The Trade Desk’s recent clean room data integration with Disney is a good example.
Using the Disney clean room product – which is backed by Snowflake, Habu and InfoSum – advertisers that buy through The Trade Desk can target audiences in CTV households by matching to Disney’s identity graph. Advertiser can use Unified ID 2.0 IDs (programmatic web IDs based on emails and phone numbers) to target those audiences across Disney properties.
‘Mud rooms’
But the seemingly clear distinction between walled garden and open data clean rooms is actually a muddier spectrum in the real world, according to Therran Oliphant, SVP of data and technology for the programmatic agency Essence.
For example, The Trade Desk may have an integration with Disney’s clean room, but Disney cycles its UID2s every month. That means a big TV advertiser running a campaign for a few weeks will be able to frequency cap during the course of that campaign, but won’t be able to create durable profiles based on the IDs. By the following month, those UID2s will register as new audiences.
Still, the Disney/Trade Desk model is actually more open than other clean rooms in the market.
The Trade Desk’s ad buying integration with Walmart is another clean room-based product. Advertisers can use Walmart data to target Walmart customers and attribute to Walmart sales. But that’s more like a walled garden, since the Walmart campaign IDs can’t be associated with open web programmatic IDs. (Disney’s month-long window is generous by comparison.)
Similarly, Roku and Kroger have a private marketplace partnership to attribute Roku ad sales to Kroger store sales. That’s a cloud-based integration that will soon become part of the Roku clean room product (which is built on Snowflake), Roku’s VP of product management, Louqman Parampath, told AdExchanger earlier this year.
Over the wall
But walled garden-style clean rooms are not the only course.
Data clean rooms began with walled gardens because the walled gardens needed them. Recall that Google’s ad server would be illegal under GDPR if it leaked ad IDs. And the other first movers – broadcasters (like NBCU) and retailers (like Walmart) – are incentivized to create clean rooms because they have large first-party data sets.
The goal for companies that are banking on clean room growth is that more and more of the market will want – and need to – participate.
LiveRamp’s Safe Haven launched last year and by Q1 this year already accounted for more than a fifth of the company’s revenue. Safe Haven’s growth is primarily driven by retail, CEO Scott Howe recently told investors. Behind every retail media network is a data clean room.
But companies such as Habu, InfoSum and LiveRamp are betting that clean rooms will become a default programmatic practice.
For example, Acrisure, an insurance company, just bought the naming rights to the NFL stadium of the Pittsburgh Steelers, which has been called Heinz stadium for more than 20 years. If Acrisure wants to attribute sponsorship value, it might do a data clean room integration with Ticketmaster, say, or maybe even with CPG company Kraft-Heinz itself to see if Heinz Stadium fans got behind the change.
After all, Kraft-Heinz and other consumer brands that previously never owned much first-party data of their own are now doing whatever they can to get it, specifically to take advantage of products like clean rooms. (Take it from Conagra).
But will it be worth the effort?
The programmatic industry has seen many promising trends with lofty goals that have yet to pan out, including multi-touch attribution and blockchain-based ad tech, to name just a few.
Data clean rooms may join that category of hopefuls – or they may really be the next big thing.
