Privacy + Cohorts

This week Google announced the “developer origin trials” of one of their new privacy-oriented, audience-targeting technologies called FLoCs.

Since you are reading this, you probably already know about Google’s privacy sandbox and interest cohort targeting, so let’s just focus on the new news.

First, the privacy protection.

A post on the Chrome blog adds some structure to how FLoCs protect privacy. There are three main dimensions.

1. Users are “part of a crowd" based on browsing similarities, not who they are individually.

2. The technology is cookieless, and no browsing data are shared with Google or anyone.

3. No cohorts are created around sensitive topics like medical or political or religious browsing.

Next, how it works.

A new FAQ on Google’s Web.dev site explains the basics, including an example illustrating each of the key roles in the supply chain (the FLoC Service, the Browser, the Advertiser, the Publisher and the Ad Tech Provider).

The FAQ is super-clear on how the FLoC service enables the browser to work out its cohort using “cohort space” and the FLoC algorithm.

But it is a little fuzzy on how large cohorts can get or how many of them there will be. Google sort-of answers this by questioning whether cohorts should be thought of as “another way of categorizing people online” or something else (their emphasis is on activities, not individuals).

The API is super-simple. When a call is made to the FLoC service only two pieces of information are returned the Cohort ID and the Browser Version. The implication is that nothing user-specific leaves the device.

Publishers can opt-out of being cohort-ized, but cohort calculations are switched on by default.

So that’s all for now.

A few links if you want to read more …

Chrome blog post. https://blog.google/products/chrome/privacy-sustainability-and-the-importance-of-and/

Web.dev article. https://web.dev/floc/#floc-algorithm

Prohibited interest categories. https://support.google.com/adspolicy/answer/143465?hl=en