Habu on the Opportunities and Challenges of Data Clean Rooms

Data clean rooms are having a moment. Fresh anti-tracking rules by tech giants and privacy regulations from California to China are forcing publishers, advertisers, and intermediaries to develop privacy-safe ways to share consensually obtained consumer data. That’s precisely the value proposition of this hot new technology.

But clean rooms are not a magic bullet for privacy compliance. Just because a customer or user willfully hands over data does not mean the data can be shared with third parties. I checked in with David Danziger, SVP of partnerships at Habu, to explore the opportunities and challenges of data clean rooms.

What forces are generating a lot of buzz right now about data clean rooms?

There are three forces that are leading to all the buzz about data clean rooms. 

The first is one that’s been building for a long time: the never-ending “craving” for more and higher-quality data. This is true for brands who seek unique data to make their marketing more effective as well as for publishers/media companies wanting data that can help make their properties as effective as possible so that they’re attractive for advertisers. 

The second force is the increasing emphasis on privacy that’s pushing brands to protect their own first party data, pushing regulators to curtail third-party data sales in the absence of clear consent, and pushing the deprecation of third-party cookies.

The third force is the desire by all parties involved in marketing and advertising to distill the enormous amounts of data, often from disparate sources, into insights that drive meaningful use cases.  

These three forces combine in ways that encourage the development of solutions in which:

• Data from a wide variety of sources can be used for collaboration without movement of the data
• The data used for collaboration purposes isn’t exposed to others in its raw or source state  
• The resulting “collaborative”’ data asset can be used to generate sophisticated insights to drive activation, analytics, measurement, and more 

These characteristics make up the core functionality for data clean rooms and the software that makes these work.

How do data clean rooms help marketers navigate privacy challenges?

Data clean rooms help marketers navigate privacy challenges in a few key ways. First, because advanced clean room software is premised on non-movement or minimized movement of data, there’s reduced chance of transmission risk. The name “clean room” is a bit of a misnomer in the sense that it implies this centralized “place” where the data all comes in but doesn’t really go out. This is somewhat the case with first-generation clean rooms and walled garden clean rooms. In reality now, the most advanced data clean rooms are really software that enables the use of data where it exists but only for specific, permissioned uses that don’t expose the source data.

And that’s the second key way data clean rooms help with privacy challenges. The source data itself is not exposed to other parties in raw form. It’s transformed for processing and for results, but no parties in the clean room collaboration see any other party’s source data in any exposed, usable form. 

The third way data clean rooms help is by creating a mechanism whereby brands can collaborate in a privacy-compliant manner with other brands and their data. They can also leverage third-party data in a “cleaner” way too.  So it’s likely that more and better data will ultimately be available to marketers for specific purposes, and only in the protected software domains of clean rooms.

Is it possible marketers and publishers could still use data clean rooms in ways that do not respect consumer consent? What I mean is this: It’s great if a company gets consent from a user/customer to collect their data, but often, companies then share that data with lots of companies without the user/customer’s knowledge or consent. How do clean rooms tackle this?

This is an excellent point and it’s correct that there’s still the possibility for “abuse.” Data clean room software is not a magic bullet for enabling compliance with consumer wishes. It does, however, make it much easier for companies that wish to exercise good data stewardship to do so. 

High-quality data clean room software incorporates powerful data governance tools to specify by which connections / partners the processed data may be used and for what purpose(s). Anything outside of those parameters would be blocked. And the software would also facilitate “masking” mechanisms such as noise injection or minimums for audience size analysis.

So your question is spot on in the sense that any company that is not careful about getting consumer consent or not serious about respecting consumer wishes isn’t going to automatically be “all set” by use of clean room software.  Rather, it would be correct to say that data clean room software is a very useful tool for companies that wish to adhere to consumer preferences because the software provides the mechanisms for data protection according to permissioned use cases.

What are the challenges clean rooms face (cultural or technical)? What problems are you hoping to solve this year?

Clean rooms face a number of interesting challenges that will become opportunities over the course of the next year or two. The first relates to what we’ll call adoption critical mass. Data clean room software will become more and more useful as more brands get comfortable using it. They’ll become familiar with the mechanisms and models for data collaboration. So as the market moves from early adoption stage toward heavier mainstream use, the use cases themselves will multiply, making the technology even more attractive.  

The second challenge is one we’ll call the “universalist” challenge. Think of this as the challenge of getting useful insights on a universal basis out of multiple walled-garden clean rooms. Consider a scenario where a major brand is trying to assess its reach among its most desirable segments. If it’s spending in Google’s DV360, Amazon’s Ad Platform, and Facebook, it can use each of these companies’ respective walled garden clean rooms (Ads Data Hub, Amazon Marketing Cloud, and Facebook Advanced Analytics). Each will give answers using the brand’s first-party data and the data out of its own clean room. But ‘normalizing’ results in a universal manner across the clean rooms in an easy, automated way is a challenge that’s still being worked through.

A third challenge is what we’d call insights democratization. Putting it simply, a lot of data clean room software hasn’t necessarily been easy for non-data scientists to use. It has required a level of sophistication that has limited adoption and blocked business users from the potential benefits of the software. As data clean room software becomes easier to use (better UIs; robust libraries of pre-defined queries; basic editing/checking tools and/or natural language approaches for generating new queries; APIs to embed clean room functionality within existing workflows) while still maintaining and expanding functionality for sophisticated users, the benefits of data clean rooms will be democratized across many functional areas. This, in turn, will lead to better adoption (see challenge one)!