Data cleanrooms do not typically integrate identity management and targeting features into the process. One process follows the other.
But that is changing. Soon advertisers will have a tool that allows them to use cookieless targeting on their websites.
“Think of it as a private marketplace that meets a cleanroom,” said Nancy Marzouk, MediaWallah chief executive officer. “Too many large brands do not want to put their data into a cleanroom because they will not copy and paste all their personally identifiable data into any platform outside of their firewall.”
They are typically a separate offering, according to Nancy Marzouk, MediaWallah chief executive officer. She says the company is testing a combined solution with a couple of large, global brands.
The difference is that the data gets encrypted into the server behind the company’s firewall. “We get the encrypted data and synthesize it to match it,” she says. “This way no personally identifiable data leaves a company’s firewall.”
Data cleanrooms like those from Google, Facebook, and Amazon can share aggregated rather than customer-level data, from more than one advertiser. Some publishers experiment with advertisers that have complementary products, such as socks and footwear.
MediaWallah provides the publisher and advertiser with a piece of code that they run where they store the data. She calls this process a multi-computation service to encrypt the data, with a mathematical algorithm that generates an encryption key on the fly.
The mathematical algorithm allows someone to look at the data and key variables in the string to match the data. The data cannot be unencrypted because of the way it passes through the multi-computation service process.
“It’s just as secure as a data cleanroom. The only thing is, the data never leaves the owner’s environment,” Marzouk says.
Marzouk expects the service to roll out during the fourth quarter of 2021. The company is testing the process with “hundreds of millions of rows of data” that get encrypted in milliseconds.
Companies do not need to move the data, and there is an identity layer built into the process to get better match rates.