Google calls Topics an evolution of their previous proposal for interest-based targeting which was called FLoC.
What Are Topics?
Under the proposal, as someone browses the internet, Chrome will keep a record of the categories of websites they visit e.g. News. Each website can have up to three categories.
Each week Chrome decides the five categories the user is most interested in and then adds a sixth at random to boost user privacy. Chrome will store these six interests for three weeks. For the purposes of initial testing, there are 350 interest groups in total.
In terms of using this information to power advertising, when a user visits a publisher’s website the companies that the publisher uses to monetize the page can access the browser’s interest groups via an API. Chrome will return up to three topics, with one chosen at random from each of the previous three weeks. Advertisers can then decide if they want to show an advert to someone with these interests and bid accordingly.
Topics will replace FLoC, which was a proposal also designed to replace third-party cookies in interest-based advertising. Unlike Topics, FLoC did not have predefined interest groups, instead, Chrome created Cohorts of browsers with similar browsing habits using Federated Learning. During the origin trial, there were more than thirty thousand groups and Ad Tech firms were expected to calculate what they thought a Cohort was most interested in.
Reactions To The Proposal
Because of this granularity, FLoC was potentially poor from a user privacy perspective. This was because the sheer number of Cohorts when combined with other information meant in theory it could be used to track users on an individual basis.
Topics has fewer larger groups that do not reveal any sensitive information to resolve both these challenges, so from this perspective, the feedback has been positive, except amongst those who take a hardline view that any tracking is unacceptable.
But some in the industry see this “broadness” of the interest groups as a downside and have questioned whether they will be too broad to be useful.
Google said they want feedback from the industry and do not want to own the final list.
Others think the current number is probably about right for these reasons:
- Topics has more than twice the number of Interest Groups as Google uses in its own advertising platforms and these are already widely used by advertisers large and small.
- If Google were to expand this to the IAB Classification it would impact most publishers’ ability to sell their audiences, and in particular, would have a negative impact on niche publishers who can command a premium because advertisers are willing to pay a premium to reach the valuable audiences that visit their content.
- There is already provision in other Privacy Sandbox proposals for ad tech businesses to create more granular segmentation. FLEDGE as I’ve commented here previously has applications that could stretch beyond remarketing.
Topics will begin testing H1 2022, with many anticipating an origin trial (first test) towards the end of Q1. It will evolve over time and for those interested in getting more closely involved and shaping what happens feedback can be posed and questions raised here.
For publishers seeking more direct participation in the tests themselves, now is probably as good a moment as any to reach out to your ad tech partners and ask how to get involved.
The first transition period for these initiatives is scheduled to begin at the end of this year.