Google is in the midst of a long-term phase-out of the third-party tracking cookies that are at the core of current digital advertising methods. Earlier this year, the company announced that the Chrome browser would no longer support them within two years. However, given that interest-based advertising is a critical component of Google’s revenue it is clear that the company is not abandoning ad tracking. What has been less clear is what exactly Google plans to replace the cookie with. Some new details have been emerging as Google has stepped up testing of its Federated Learning of Cohorts (FLoC) initiative, which it calls a “privacy first” approach to delivering targeted advertising to relevant end users.
In broad strokes, FLoC will implement a cohort-based approach to replace the tracking of individuals. The end user’s interests will still be measured in a similar way, but they will be assigned to a cohort of similar combinations by a clustering algorithm rather than being tagged with a unique identifier. The ads industry will pitch their wares to these more generalized cohort IDs rather than individual users. Google is poised to expand direct testing of this program with advertisers in Q2 of this year, but the tech giant is already promising that this new “privacy first” approach will generate about 95% of the conversions that are seen with current methods.
FLoC testing scheduled to begin with next Chrome release
Worth almost $147 billion dollars last year, Google’s ad revenue makes up the vast majority of its incoming funds. But public attitudes (and appetite for government regulation) have been shifting to be more unfavorable of personalized ad tracking. That, and similar moves by contemporaries and competitors such as Apple (which rolled out major privacy changes involving interest-based advertising with the recent release of iOS 14), are likely what have prompted Google to make voluntary reforms under the “privacy first” banner.
Google first proposed FLoC as a part of its Privacy Sandbox initiative in 2020. It looks to revamp interest-based advertising with an added layer of anonymity for the end user without significantly degrading the effectiveness of targeted ads. The fundamental approach stays the same; as end users browse the web, Google is taking note of what they interact with and how long they spend in various places. This information builds a basic profile of user interest to indicate what ads might be relevant to them. Independent advertisers place bids with Google for display of their ads to particular demographic groups that are of interest to them. What FLoC changes is the isolation and profiling of individual users. As the end user browses, Google flags them for inclusion in cohort groups based on their activity. The advertiser is now bidding to display ads to these cohort groups.
Google says that this “hides individuals in the crowd.” An added step that the search giant has proposed is to use on-device processing to keep web history private; presumably this means that Google will generate the cohort ID in the local Chrome installation before passing it on to the server.
Of course, the big question for the interest-based advertising community is whether or not this will work as well as the current system. Google says that its internal testing of Google Audiences indicates that it will be 95% as effective in terms of conversions per marketing dollar spent; not perfect, but not likely to send advertisers running for markedly inferior alternatives.
Google has published a FLoC whitepaper that advertisers can use to run their own tests. It has announced that it is facilitating further testing with the next release of Chrome, which will provide access to FLoC-based cohorts and allow advertisers to run origin trials. Given that there is generally a new Chrome version about once every six weeks, this is expected to be available sometime in late March or early April.
Will interest-based advertising substantially change?
It remains to be seen exactly how much first-party data Google will restrict itself to collecting, but FLoC and other items in the “privacy first” proposals would do a great deal to shield end users from exploitation by third-party cookies that often form frighteningly detailed profiles. Google has also formally committed to never again build individual-level tracking identifiers as part of its “privacy first” pledge.
The most promising development for end users would be if web browsing history and activity is indeed restricted to the local browser. While Google might still grant itself some sort of access to locally-stored interest-based advertising data via Chrome, it would be beyond the reach of third-party data brokers that scrape up every piece of personal information that can be found and assemble invasive detailed profiles that are sold freely. It would also greatly reduce personal information leakage by way of data breaches.
There are some limitations to the FLoC proposal. One is that it will only be available for users of the Chrome browser. Some in the interest-based advertising industry speculate that this, combined with other major browsers simply blocking cookies without offering any ad tracking alternatives, could create a spike in the attempt to use device fingerprinting techniques to continue with business as usual. Sites that have users log in might also be able to connect those accounts directly to the cohorts they’ve been placed in.
“Privacy first” is on trend, but the market may not be buying it
The “privacy first” campaign and the Sandbox proposals are saying what end users (and regulators) want to hear, but are also already receiving pushback from both sides.
Marketers in the interest-based advertising field are generally not happy with the privacy first approach, as despite Google’s claims of near-parity with current methods it nevertheless represents a downgrade and extra restrictions. Some ad tech companies call it anticompetitive, and the United Kingdom’s Competition and Markets Authority has already opened an investigation of this nature in response to complaints from an industry lobbying group.Google’s #Privacy First campaign and Sandbox proposals are saying what end users (and regulators) want to hear, but are also already receiving pushback from both sides. #respectdataClick to Tweet
On the opposite end of the field, the Electronic Frontier Foundation (EFF) has characterized the FLoC proposal as a “behavioral credit score” and worries that it will lead to discrimination. The group also expresses concern about how secure the cohort IDs will be. The initial proposal indicates that FLoC may share even more information about browsing history than the current system does, putting users at greater risk if they are somehow paired with their cohort ID by a third party.