Nielsen believes it has the answer for a cookieless world via a complex workaround that uses machine learning across numerous context clues to help track and categorize people even when they aren’t logged into brand or publisher databases.

In an announcement today, the company said its approach will make Nielsen One the only platform to validate brand or publisher first-party server data using real consumer behavior. That ultimately could have a huge impact on media measurement given that Nielsen Digital Ad Ratings currently measure nearly 90% of digital video spending in the U.S. Nielsen says 60% of that measurement is already resilient against the impending loss of third-party identifiers in Google Chrome and elsewhere, and that the steps it’s announcing will make it fully resilient.

To measure “authenticated” traffic—where people are logged in to a site or otherwise give their consent—Nielsen will use all available identifiers and first-party data from clients, including anonymized e-mail addresses, custom demographic segments and Unified ID 2.0, the industry-wide identification system embraced by The Trade Desk, Neustar and publisher groups. This will ensure Nielsen measurement can work within and across walled gardens and reduce reliance on other third parties, the company says.

To measure “unauthenticated” traffic—people who aren’t logged into a site and where no alternative identifiers are available—Nielsen says it’s developed a machine-learning technique using contextual data signals. These include time of day, browser, content, device information and Google’s Federated Learning of Cohorts [FLoC] groups to help identify and categorize people. This machine learning modeling will be validated against Nielsen panels “for representation and accuracy,” the company says.

Nielsen will participate in a pilot with the Association of National Advertisers and World Federation of Advertisers on its solution, but hasn't altered its timetable because of the extra time afforded by Google’s decision last month to delay the demise of third-party cookies in Chrome until 2023, says Mainak Mazumdar, Nielsen’s chief digital officer. The ANA last month also launched a pilot with Comscore to test media measurement without cookies.

Nielsen will also participate in Google’s testing of FLoC, Mazumdar says. “We are working with Google and all the other walled platforms to make sure that our measurement is interoperable.”

Simplifying Nielsen’s job somewhat is that it’s not looking to replace ad targeting to individuals, but to measure unduplicated audience reach across demographics and cap frequency of ad impressions for people across digital and linear video.

“This is purely a measurement use case,” Mazumdar says. “We want to know are they real people or bots or spiders? Are there persons behind these devices? Who they are, and are they watching the same ad across multiple devices? That’s very important. It reduces waste. It makes it more efficient.”

Unlike cookies, which Nielsen calls “brittle identifiers,” the Nielsen ID System will use identifiers that remain consistent over time and are tied, anonymously, to actual people and households, the company says. It will all be built, Nielsen says, on “proprietary confidential computing technologies” to safeguard privacy.

Nielsen’s ID graph has the advantage of being the only one validated against Nielsen panels, Mazumdar says.

“The focus right now is to make sure the panel is balanced and representative,” he says. “The focus is to make sure that when we use this panel to calibrate big data, to validate who is behind all these devices, that it’s representative of all ethnic groups, race and gender.”

Nielsen has been working with the Media Rating Council from the outset of developing its ID System, Mazumdar says. “It’s under audit right now. The timeline is going to be continued when this gets activated in the marketplace, but we are 100% committed to doing it.”