• by Laurie Sullivan  @lauriesullivan, October 22, 2020

Google has released findings of a preliminary study that tests the use of audience cohorts, people with similar browsing histories and interests who might be targeted collectively rather than individually. It almost seems like a step backward by stepping sideways to achieve its goal. 

The test aims to support the end of individual cookie targeting and cross-site tracking.

Using its Federated Learning of Cohorts API, a privacy-preserving mechanism in its Chrome Privacy Sandbox, Google managed to show that targeting interest-based cohorts perform better than random user groupings.

Earlier this year, Chrome published an intention to implement a Federated Learning of Cohorts API with a goal to preserve interest-based advertising.

The idea is that grouping people into audiences protects their privacy. The goal of the paper is to share initial results about the efficiency of certain algorithms. Google researchers believe that in the long run, the tech industry can use cohort IDs in their ads to personalize algorithms.

The report discusses different algorithmic approaches and thoughts related to data privacy. For example, the report elaborated on a privacy vs. utility trade-off.

“The more users share a cohort id, the harder it is to use this signal to derive individual user’s behavior from across the web,” according to the report. “On the other hand, a large cohort is more likely to have a diverse set of users, this making it harder to use this information for fine-grained ads personalization purposes.”

Google announced the end of third-party cookie support in Chrome in 2022.

At the time, the plan was to start the first trials by the end of 2020, starting with conversion measurement and following with personalization. This could provide Google entrance into personalization using something other than cookies through its open standard initiative, Privacy Sandbox.