Google only launched the Topics API in January, and opened it to global testing in July. The Topics API—which stands for application programming interface—replaced an earlier proposal that was found to have privacy holes, called Federating Learning of Cohorts or FLoC. Google has been very deliberate about how it changes data-sharing on Chrome web browsers and Android devices. It is taking more time than Apple, which shook the ad world by kicking cookies off Safari beginning in 2017, and last year started preventing apps from gathering too much information from consumers on iPhones, which hurt their advertising businesses.
Unlike Apple, Google is the world’s largest internet ad company, and if it were to withhold data from rivals in ways that harm their advertising businesses, it would raise even more scrutiny. Google faces a pending anti-trust case from the U.S. Department of Justice, which promises to expose more details about how the search giant controls large portions of online advertising on the publishing and advertising side. Earlier this year, Google postponed the deprecation of cookies on Chrome another year, until 2024, claiming it was a reprieve for the industry to test programs like Topics and other alternative IDs.
There are plenty of skeptics of Google’s Privacy Sandbox proposals. “The only company that really knows how this affects the economics of the web and the media industry is Google,” said Jason Kint, CEO of Digital Content Next, a publishing industry trade group. “That’s the real problem here.” Kint is an outspoken critic of tech giants such as Google and Meta, pushing for due compensation for publishers.
Related: Google's cookie confusion is splitting the ad industry
Where Google Topics API received higher marks was in the area of transparency, Farrow said. The proposal gives consumers more details about why they were tagged with certain interests, based on their browsing, as opposed to the older FLoC proposal. FLoC revealed less about what categories people were being lumped into as parts of the cohorts they were assigned. Topics are more contextual, while FLoC aimed for anonymity by diluting individual identities in larger pools of audiences. Still, FLoC raised concerns that nosy participants in ad markets might be able to reverse engineer identities to obtain personal information.
Xandr’s initial tests into the Topics API found some strengths in how Google organizes categories that can be used to target ads. “We saw that topics were broadly in line with what we expect for users that we know,” Farrow said. Xandr studied how Google analyzes content from websites and how it labels the sites for ad targets.
Google is using the context of websites such as the New York Times, Vogue Business, Elle and others, to generate the interest categories. Google Chrome shares a small sample of interests, collected within a limited timeframe with advertisers, instead of using third-party cookies that reveal more personal information about the consumer. Publishers and brands are working on their own identity frameworks that would keep personalized advertising going, based on getting consent from the consumer and first-party data relationships. There are other Google ad managing tools for publishers, which help them use those alternative methods.
Xandr’s test also found that Facebook.com is one of the domains that inform the ad traits on Chrome. The website contributes interests such as news, social media, sports, arts, literature, entertainment and home and garden, among others. There are about 350 topics. Xandr found scale was an issue, since 12.4% of web domains in the U.S. were not assigned topics. “The proportion of domains without topics is not negligible,” Xandr wrote in a blog post in July, outlining some of Topics tests.
Xandr is not ready to go full-steam ahead into adopting Google’s Privacy Sandbox proposals, Farrow said. Advertisers and publishers are still cautious: “Does this actually provide value or is this just a thing that is not really necessary?” Farrow said.