After delays and missteps, Google is outlining its latest plan for advertising in Chrome web browser following the demise of the cookie.
The tech behemoth is now ready to test what it calls Topics advertising, an iteration on an experiment that it ran last year, where it can serve personalized ads to Chrome web browsers to groups of people, or cohorts, so ads do not target individuals.
This comes after it received push back on its previous test of its Federated Learning of Cohorts [FLOC] model, which hid users in pools of anonymity. But that solution was found to allow for bad actors who could use techniques to trace “cohort” data back to personally identifiable information.
Google is experimenting with these targeting techniques in order to phase out third-party cookies by late 2023. Cookies have long been criticized as invasive tracking devices, collecting data on people’s internet behavior without their consent.
Google's latest cookie replacement has some privacy leaders still concerned that Google can continue to keep tabs on web users. Meanwhile, the ad tech and ad targeting crowd are more concerned that Google’s new limits on data could stifle robust targeting that brands use to message consumers.
Instead of FLOC, Google plans to replace cookies by sharing a small sampling of topics—interests collected about users as they traverse the internet on Chrome—with websites.
Google will start the test with about 300 interest topics, like fitness and travel, a number that could swell into the low thousands, according to Ben Galbraith, product director at Chrome, who held a conference call with media on Monday to explain the ad targeting experiment. Google will collect five interests on a user each week, for up to three weeks, and then expose three interests, one from each week, to websites they visit.
“When you visit a participating website, one topic of these five, from each of the past three weeks is shared with the site, which can include its advertising partners, which then can be used to display ads that will be the most relevant,” Galbraith said. “And let’s contrast this for a moment with cookies, where advertisers track users and build profiles of them to understand what those topics of interest might be.”
Galbraith said that Topics “improves user privacy” from FLOC. For one, the process exchanges fewer data points. Also, FLOC was run by machine-learning, and using artificial intelligence was an opaque way of categorizing consumers and their interests. Topics are simpler to understand, and consumers can see exactly what topics they fall under, and control them, Galbraith said. “The design of Topics was informed by our learnings from our earlier FLOC trials,” Galbraith said, “which included lots of great feedback from the community and as such Topics replaces our FLOC proposal.”
A vocal Google critic, Pete Snyder, the director of privacy at competing browser Brave, said that Google’s topics proposal falls short of solving key privacy issues. Brave, a browser with 50 million monthly users, avoided Google FLOC trials, preventing it from working on the browser, and plans to stay out of the topics experiment, Snyder said. One of the problems, according to Snyder, is that Google would still collect data on consumers.
“The core idea here, the main central concern here, to put it bluntly,” Snyder said, “who is Google to tell me what I think is sensitive about myself?”
Google will restrict topics categories in some clearly sensitive areas, like racial affinities and sexual orientation, but Snyder said that still might not cover people’s full range of privacy preferences.
Google will allow consumers to opt-out of interest topics assigned to their browsers, or opt-out altogether, Galbraith said.
Meanwhile, others said that Google’s new proposal seems like it might not go far enough to enable personalized advertising to flourish online. Tal Chalozin, CEO of Innovid, a connected TV ad and measurement platform, said that Chrome Topics seem to be an even less ambitious alternative to FLOC.
“It’s clearly a complete slim-down of the grand vision of FLOC,” Chalozin said, “which was significantly more robust and a broader solution, but it seems like it got a lot of heat from the industry.”
Mike Woosley, chief operating officer of Lotame, an ad tech and data company, had similar issues. “Google describes ‘Topics’ as a browser-side utility that will assign a user up to several from a ‘handful’ of interest areas that will persist for three weeks,” Woosley said by email to Ad Age. “This type of capability hearkens to contextual advertising techniques circa 2005. Unfortunately, the technology as described would be grossly insufficient for the needs of the vast majority of modern marketers who require detailed personas to determine marketing voice, segment customers, measure brand affinity, and tune marketing for complex products like insurance with very detailed segmentations.”
Google is attempting to catch up to other web browsers and device companies, like Apple, which shut down third-party cookies in Safari in 2020. Also, there has been more regulations in Europe and the United States, which have cracked down on programmatic advertising tactics that rely on sharing data without people’s permission.
Google is the largest internet advertising company, delivering ads to millions of websites that use its ad network. Google also makes Android devices, the Chrome browser, and owns popular platforms like YouTube and search. That powerful position enhances the scrutiny any time it changes how its ad platform works, which explains the long timeline to phase out third-party cookies. On Monday, Galbraith said that Google still intends to oust third-party cookies by the forecasted end of 2023.
Third-party cookies have been an especially tough problem for Google, because it also relies on them to target ads within its ad platform. There were antitrust concerns that Google could benefit by blocking third-party cookies. The fear was that Google could shut competitors out of the ad equation, while it still sat atop all the data coming from Chrome to still effectively target ads within its platform.
Those concerns, were partly why Google redrew its plans. Last year, Google delayed the deprecation of third-party cookies from 2021 to 2023, and it gave assurances to European regulators that its ad platform would operate under the same data restrictions as competitors.
On Monday, Galbraith pointed to the commitment Google made with the U.K.’s Competitions and Market Authority. “We will not self-preference our own advertising business,” Galbraith said.
Google has only begun to test topics-based ad targeting, and is looking for ad tech companies, browsers and brands to plug into its software to try it. Google uncovered a series of flaws with the original FLOC proposal when it went into the trial phase last year. It turned out that other companies in the ecosystem could access enough data in FLOC to uncover interests on individual consumers, through a tactic known as “fingerprinting.” Fingerprinting is when a data or ad tech company has the ability to read settings on a person’s device in order to pinpoint who they are and when that person visits another website, or returns to a website, they can be better targeted with ads.
Google claimed the Topics ad platform would limit the ability to fingerprint consumers. It’s not that fingerprinting is impossible, but Google’s hope is that with fewer data points, there are fewer signals for outside companies to latch on to pinpoint the end user.
Brave’s Snyder said that the topics experiment seemed to be a rebranding of FLOC with few meaningful changes. “Topics are not any different than cohorts,” Snyder said. “You’re still grouping people into a small number of buckets, or maybe I should say a large number of buckets.”
Chalozin said that the biggest problem with Google’s proposals, so far, is that while topics could solve ad targeting, brands and ad companies need better measurement and attribution fixes. As the web becomes more anonymous, it becomes harder for brands to follow when ads were seen by consumers, and when those ads were acted upon. “You want to go back to see the ads that led someone to put something in their shopping cart,” Chalozin said.
Galbraith said Google is still working on measurement and attribution in its broader experiments within its so-called Privacy Sandbox, of which topics targeting is only one aspect.
Sara Stevens, VP of digital capabilities at Epsilon, the data company owned by Publicis, said that the entire industry is already adjusted to ad targeting without cookies by creating new tools for Apple’s Safari web browser. Brands are working with companies like Epsilon to tie consumer data points and targeting ads through alternative identifiers.
“It’s good that Google went back to the drawing board,” Stevens said. “They’re staying true to their desire to adhere to user privacy, but the devil is in the details. So let the testing begin.”