Data clean rooms could be the perfect technology for the privacy-first era

data
(Image credit: Shutterstock)

In the post-cookies world, companies will need to look for equally effective ways to collect, share and analyze data for the purposes of marketing, without compromising on privacy.

One solution is the data clean room, a service that allows brands to share data and insights with partner organizations, while still safeguarding the privacy of those whose data is shared.

To find out more, TechRadar Pro spoke to Barak Witkowski, EVP Product at marketing analytics company AppsFlyer, which recently launched its own data clean room in partnership with Intel

He explained precisely how these services work and offered insight into the techniques and technologies that will allow marketers to remain compliant with data privacy regulations in the years to come.

What is a data clean room?

A data clean room is a piece of software that enables brands and their partners to share data and gain mutual insights while fully preserving the privacy of the users, by not sharing any personally identifiable information or raw data with one another. Think of it as Switzerland for data.

Currently, there are two types of data clean room solutions available in the martech industry: walled gardens solutions and independent solutions, both of which carry their own advantages and limitations.

What are the advantages for consumers, beyond the compliance benefits for brands?

Consumers have grown accustomed to a certain level of user experience when interacting with brands - such as seeing personalized, relevant content within an app - which has traditionally been facilitated by access to user-level data, such as cookies on the web or identifiers on mobile devices.

However, exchanging user-level data simply because it is what these insights are built on has created the privacy problem that exists today. Consumers want to know how their data is being shared (rightfully so), and new privacy regulations reflect this change.

With AppsFlyer’s Privacy Cloud and other data clean room solutions, consumers can still get the great value and experience they expect from brands, without any privacy concerns as to how their data is being used. 

There should be no compromise between customer experience and privacy. Any compromise on either side is a lose-lose situation, most importantly for the end user.

With third-party cookies on their way out, what role will clean rooms play in supporting the demand for both privacy and marketing insights?

Data exchange (between brands and partners) has always been the basis for accurate and actionable measurement, which enables both sides to grow their businesses and give better experience to the end users. Up until now, however, this data exchange has been done based on user-level data only. 

Data clean rooms are a solution that maintains the great value and customer experience currently enabled by cookies, identifiers, and other user-level data without the privacy concerns. 

They allow brands, app developers, ad networks, and so on to collaborate, share data, and generate insights, all of which is essential to creating a great experience, while still preserving user privacy. Brands will be able to define their own business logic, compliance and data governance, to ensure that sensitive, user-level data is kept safe and private - exactly how it should be.

Are there use cases for data clean rooms outside the marketing industry?

Data clean rooms are used in various industries. In the most general way, they are secure environments where multiple parties can collaborate on sensitive and restricted data sets. Applications of this concept can be found in healthcare and life sciences, insurance, fintech and other domains where sensitive data such as personal identifiable information (PII) has to be shared between multiple parties to perform analyses and generate insights.

How does AppsFlyer's solution differ from others on the market?

AppsFlyer’s Privacy Cloud allows our customers and partners to keep up and comply with all the various privacy regulations and guidelines, while still getting the accurate insights they need in order to grow their business.

Other existing data clean rooms have certain limitations. For example, data clean rooms from walled gardens have no cross-channel access, meaning first party data is mostly shared with their own data sets. Other independent data clean room solutions may be limited to first party data granularity and have small partner ecosystems. But mostly, they lack the expertise of generating insights that the marketer needs, which is what AppsFlyer has been doing successfully for the last decade.

The AppsFlyer Privacy Cloud is cross-channel and offers best in class measurement features needed by the marketer, access to the ecosystem of AppsFlyer’s integrated partners, aggregated reporting, and is suited for both business users and marketers.

As part of this, we have entered into a long-term collaboration with Intel to work on privacy-preserving cryptographic solutions such as homomorphic encryption (HE) and private set intersection (PSI).

What's the significance of homomorphic encryption in this context?

Homomorphic encryption makes it possible to generate aggregated insights about the encrypted data, without ever decrypting it. It remains fully encrypted all the time. Hence, it’s a ‘zero trust’ technique, in the manner that even the operator of the data clean room doesn’t have access to the plain data.

Just like other forms of encryption, homomorphic encryption uses a public key to encrypt the data. Unlike other forms of encryption, it uses an algebraic system to allow functions to be performed on the data while it’s encrypted. Then only the individual with the matching private key can access the unencrypted data after the functions and manipulation are complete. This allows the data to be and remain secure and private even when someone is using it.

Tell us about the collaboration with Intel

Our collaboration with Intel is about more than just utilizing the company’s hardware, it's Intel’s full PSI/HE stack on top of its hardware solution, and will allow us to develop and advance privacy-preserving cryptographic solutions.

We understood that with the scale of our data, the amount of computation needed to run these cryptographic solutions is simply not realistic. By partnering with a strong leader such as Intel, which shares with us the vision of innovation around privacy preserving cryptographic technologies, we will be able to more quickly develop a scalable solution for the ecosystem. 

These solutions will make the AppsFlyer Privacy Cloud more holistic, as it will give all parties the option to use “zero trust technologies”, where sensitive data doesn’t enter the cloud before being encrypted.

What other technologies does AppsFlyer see playing a role in a post-cookies marketing ecosystem?

There are two advances that we see taking on greater importance in a world without user-level identifiers: incrementality-based solutions and predictive analytics. 

With access to user-level data, marketers have traditionally been able to “match” a desired action (ad click, impression etc.) with a conversion. But as we move away from user-level data, this will become harder to do, and we'll see an increased focus on measuring incremental lift. This will allow marketers to understand the real impact of their investments, by uncovering which conversions are a result of marketing efforts, and which would have happened organically. Incrementality-based solutions use test and control groups to isolate many affected variables, and help marketers optimise accordingly. 

Predictive analytics will be particularly useful for any mobile marketer focused on iOS, but not exclusively so. Following Apple’s update to iOS 14.5, which has essentially depreciated the identifier known as IDFA, the company presented its SKAdNetwork framework. Within this framework, marketers have a very short window in which to measure campaign performance and make decisions: 24-72 hours. Predictive technologies will enable marketers to leverage early signs of engagement generated within this initial window, and predict long-term campaign performance accordingly. 

Both of these solutions are game-changing advances that will be powerful tools in any marketer’s arsenal.