Google Introduces Topics to replace the controversial Federated Learning of Cohorts

Google Introduces Topics to replace the controversial Federated Learning of Cohorts

Google has canceled plans to replace third-party cookies with their Federated Learning of Cohorts (FLoC). Instead, its new proposed model is Topics.
by Usercentrics
Feb 15, 2022
Book a demo
Learn how our consent management solution can improve privacy and user experience for your users.
Get your free data privacy audit now!

There have been rumblings for some time that third-party cookies will become extinct. Cookies are a very old technology in internet terms, and there are issues with how invasive their tracking can be. In this era of increasing concern about and regulation of online privacy, it’s not surprising that new options are being explored. Just what the right replacement for third-party cookies is remains to be determined.

 

In late January 2022, Google announced that they would not be going ahead with Federated Learning of Cohorts (FLoC), their replacement for third-party cookies announced in mid-2021.

 

See our article: Google’s Federated Learning of Cohorts: why you need to give a FLoC

 

Google announced instead that their new proposed solution is Topics, another API-driven aggregate model that would learn about users’ interests in order to target them with advertising.

How will Google Topics work?

Like FLoC, Google Topics will work on an interest-based model. As users browse online, the browser will learn about their interests from sites they visit. For now this only applies to the Chrome browser, though it has nearly 70 percent of browser market share. This browsing history will store a maximum of 300 interests — with plans to expand that over time — for three weeks, after which point they will be deleted and replaced with a newly built-up set.

 

What exact types or amount of information qualifies as an “interest” is not entirely clear, though information often categorized as “sensitive” will not be included.

 

Learn more: Personally Identifiable Information (PII) vs. Personal Data – What’s the difference?

 

With the FLoC model, users’ browser histories would have been compared and categorized into “buckets” of similar interests, instead of each user’s established topic or category interests. For example, under FLoC users might have been categorized as having similar interests because they both visited the usercentrics.com website.

 

With Google Topics, users might be categorized together by behavior, e.g. because they were reading about consent management solutions, but not because they were doing so on the same website. Topics is similar to contextual targeting, a more classic type of advertising. Privacy advocates have expressed concerns that Topics will still be informing third-party trackers what sites users visit.

 

Google says that users will be able to review and remove topics from their lists, and notes that the design of Topics was heavily influenced by feedback from FLoC trials and browser users. Data protection and privacy concerns were a big criticism of FLoC, and advertisers weren’t fans of the concept, either.

 

For users, once their topics of interest list has been built, Google categorizes subsequent sites visited based on this list. If their topic list hasn’t been built yet, or a user browses an uncategorized site, an algorithm uses the site’s domain name to estimate a topic.

How does the Google Topics API work in advertising?

Websites will be able to use a Topics API for advertising purposes, and Google’s current plan is to start trials of it at the end of the first quarter of 2022. A user’s browser will randomly select three topics from the top five of the user’s current list of interests, one from each of the three weeks of data that the browser stores. (Learn more: technical information about Topics API)

 

The website running the Topics API can then use these three topics, sharing them with advertising partners to determine which ads to display to the user. In addition to editing their interests list, Google says that users will also be able to turn off the Topics API entirely.

 

Advertisers have expressed concerns over the restrictiveness of Topics. However, Google insists advertisers will continue to have additional signals to decide on what ads to show users. Topics would be only one option. For example, data about the page the user is currently on, like the article they’re reading, can provide additional information and context.

 

The other browser vendors refused to consider adding FLoC, so while they would have the option of adding the Topics API, it remains to see if any will. There are still concerns over Topics not really solving any issues or introducing anything new, as well as concern over Google’s “indecision” over their choice of direction for privacy and technology.

 

As users, privacy advocates and advertisers have concerns over Topics, we may well see changes or yet another proposed replacement for third-party cookies from Google in the future.

 

Still have questions about data protection and privacy in browsers? We’re here to help. Talk to an expert today!

Home Resources Article Google Introduces Topics to replace the controversial Federated Learning of Cohorts

Related Articles

German Data Protection Officer calls for Google to offer “reject all” cookie consent option

German Data Protection Officer calls for Google to offer “reject all” cookie consent option

Hamburg Data Protection Officer calls for Google to fix unequal access to consent and enable a “reject all”...

How Server-side tracking and consent management creates harmony between privacy and data

How Server-side tracking and consent management creates harmony between privacy and data

Managing user consent is no longer just an in-browser discipline. Evolving technology and legislation require brands to...