On trend: data clean rooms enabling privacy-safe collaboration
For those deep in the IT supply chain, clean rooms may bring to mind state-of-the-art ‘gigafabs’ that must be protected from dust to successfully pattern thousands and thousands of silicon wafers with tiny features. But what about data clean rooms? Just as the silicon fab is a safe space for producing semiconductor chips, a data clean room is a safe space for exploring sensitive information. The concept has been around for a while, but is catching on for reasons that we’ll dive into.
In a nutshell, data clean rooms are ideal for the scenario when company A wants to explore how its data correlates with company B, but the data itself must remain private to each of the owners – A and B. They do this by focusing on the intersections, also know as an identity graph.
Advertisers can use the virtual spaces as a digital walled garden to explore whether an advertising platform is a good match for them by correlating their customer data with the platform’s user base. Critically, to preserve privacy and comply with regulations such as Europe’s General Data Protection Regulation and the California Consumer Privacy Act, sensitive elements such as personally identifiable information (PII) are stripped out.
Data clean room providers such as US firm Habu, which has offices in San Francisco and Boston and came out of stealth mode in 2020, promise secure data collaboration, controlled data access and comprehensive analytics. Driving customers towards solutions such as Habu’s and others is the decline in the effectiveness of tracking tools such as cookies, which have been hobbled by the requirement for users to ‘opt in’ – a European initiative to force website owners to be more transparent about their tracking behavior.
A new era for charting customer journeys
To piece together the customer journey in this new era, clients are taking to data clean rooms that facilitate the secure and private interaction between streams of customer and partner data. In sectors such as retail, this is allowing major supermarkets and international fast-moving consumer goods firms to build on data insights, but without handing over sensitive information. The same tools mean that media firms, for example, can make a compelling case to potential advertizers without compromising the privacy of their subscribers.
In 2021, Disney announced a collaboration with Infosum, Snowflake and Habu that allowed clients to benefit from the media giant’s proprietary audience graph (a dynamic dataset that helps advertizers find segments from pasta-lovers to ice hockey fans) within a cloud-based walled garden. Using the tool, potential advertisers can gather “a more accurate understanding of consumer engagement and outcomes,” according to reports provided at the launch of the service.
Speaking at Beet Retreat (a gathering of high-level media executives held in Santa Monica), Laura Nelson – senior vice president of ad products and enablement at Disney – gave a bit more detail on what this entails. By leveraging a shared media database, customers can compare their instance with Disney’s instance to gauge audience overlap, with the whole process taking place within the ‘safe space’ of the data clean room.
Data clean rooms are on the rise, but many unknowns remain. Disney’s Nelson points out it was important for the US firm to work with multiple partners as it beta-tested its cleanroom to bolster its understanding of what the top features should be. In turn, this knowledge then fed into the next phases of the project to help prioritize development on the platform. And it’s not just media companies that are experimenting with the concept. Financial services providers are watching keenly too as they seek to explore use cases with their customers in a digital environment that protects everyone’s data assets.
To keep everything local, data warehousing company Snowflake – which you’ll remember as being part of the Disney collaboration – now has a number of identity resolution services that run natively on its cloud platform. This means that users can fix issues in their data within their own environment without having to move the data anywhere else, which is a boon for privacy. The data from all partners is assigned identity numbers, which allows it to be compared, but (crucially) without bringing any PII into the room. And – as has hopefully now become clear – the process enables valuable insights such identifying the size of the overlap between subscribers and customers to help a potential advertiser to determine whether a publisher’s platform is the right match for them.
5 December 2023
4 December 2023
30 November 2023