Earlier this week, the Justice Department filed a lawsuit against Google, citing its dominance in the search market as anti-competitive. But beyond the big government and big tech theater, advertisers watch the countdown to the “cookie pocalypse” as Google prepares to withdraw support for third-party cookies in its Chrome browser.
This is of course due to changes in user tracking laws around the world. Following that monumental announcement in January, the Google Chrome team tacitly launched its Privacy Sandbox initiative.
This project represents Google’s attempts to figure out how online advertising will work after the industry’s universal currency devalued. And while not all suggestions were received with appreciation, the project is arguably the most realistic portal for how the $ 125 billion industry will work after 2022.
This week, Google released the results of the latest experiments conducted as part of the Sandbox Initiative. The analysis focused on the concept of replacing one-one targeting – the central premise of cookies – with cohort or FLOC targeting in order to examine the habits of similar groups of users.
“More specifically, the FLOC API is based on a cohort assignment algorithm,” the report said. “This is a feature that assigns a cohort ID to a user based on their browsing history.”
Further basics for FLOC targeting are:
- A cohort ID should prevent individual cross-location tracking
- A cohort should be made up of users with similar browsing history
- A user’s browsing history should be “hashed” to prevent identification
The accuracy of the algorithm’s targeting capabilities was tested using publicly available datasets that would summarize specific interests or industries.
The online advertising giant claimed that this process of targeting was improved by 350% and that there was a 70% improvement in targeting accuracy “at very high levels of anonymity compared to random groupings”.
The authors of the report acknowledge some of the “tradeoffs between privacy and utility” in the system. For example, the larger the cohort, the better the privacy guarantee. However, the larger the cohort, the more difficult it is to deliver a relevant ad to a user.
“Still, it is encouraging to see that a fully decentralized approach can produce cohorts that are roughly 85% of the quality of a fully centralized algorithm,” the report said.
A team effort?
Let’s say your analysis is correct and FLoCs are better than third party cookie-based targeting. Then put it on the market and let FLoC solution beat third party cookie based solutions in the market instead of blocking one tool over the other for competitive reasons. If not why
– Danny Sepulveda (@DSepDC) October 23, 2020
Google is encouraging other ad tech companies to conduct similar experiments and test their algorithms on their own proprietary datasets.
It also accepts feedback from third parties, particularly through the W3C web standards committee, where colleagues generally don’t shy away from expressing dissent, even when Google’s critics in Congress claim that it has an oversized influence there.
Some claim that Google’s results have interests of their own, despite soliciting third-party feedback. Additionally, privacy advocates have expressed their problems with such new approaches to online ad targeting.