Nielsen is cooking up new ways of measuring audiences without relying on unique trackers like third-party cookies, as those data sets are on the cusp of being phased out. The audience measurement powerhouse revealed its ID resolution system on Thursday. It uses machine learning to process data from its own panels—along with data from publishers and advertisers—to assure advertisers that when an ad is viewed by an individual, the audience is unduplicated and accurately measured across all platforms. 

Nielsen's ID resolution system builds a profile by capturing media-related signals and behavior—even across devices—and combining the data using deep learning techniques, says Mainak Mazumdar, chief data and research officer at Nielsen. The end result is essentially “a computational graph that allows us to link signals into a person,” Mainak says.

Nielsen ID “utilizes persistent, device-agnostic identifiers that have been validated against its high quality panel truth sets, enabling Nielsen to verify demographic characteristics and viewership over time,” according to a statement from the company.

Nielsen says its new system lets clients measure audiences across platforms and can reduce duplication—when an individual sees the same ad across multiple platforms. “The ID system would now be able to say how many unique people they can reach,” says Mainak.

The new system comes as the industry moves away from third-party cookies, spurred by shifts from major players like Google. “Cookies have been very useful—it has very unique advantages, and it’s delivered a lot of things for advertisers and media companies,” says Mainak. “I think the trend now is there’s going to be something different.”

Nielsen says the data provided by its ID resolution system is interoperable with other IDs including The Trade Desk's Unified ID 2.0. Mainak expects to roll the capabilities of its ID resolution system into multiple products in the future, starting with digital products like digital ad ratings and total ad ratings.