Private Aggregation filtering IDs launching on pre-release channels

Hi everyone,

Starting in M127, developers using the Private Aggregation API will be able to optionally specify a “filtering ID” when making a contribution. Filtering IDs allow for aggregation service queries to be split by a developer-specified dimension. This provides more flexibility, e.g. to process some measurements at different cadences to others. Further detail is available in the explainer.

Initially, this feature will only be available for 50% of users on pre-release channels of Chrome, starting with 50% of Canary and Dev channels on Monday 17 June 2024. This will limit any compatibility impacts to this small fraction of traffic, while still allowing for testing.

To use the new filtering ID feature, you will need to upgrade your Aggregation Service release to version 2.5 or later. When this feature is enabled for a user, their reports’ version field will use the updated value of "1.0" (instead of "0.1"), even if filtering IDs are not specified. Guidance for older versions is provided below:

• If you are using version 2.4, the new report version is supported but the filtering ID feature is not supported. Ad techs intending to use filtering IDs should update to version 2.5 so that reports are processed correctly.   

• If you are using an older Aggregation Service release, you should filter out reports with the new version number before each aggregation to avoid job failures.

If you process the "debug_cleartext_payload" field, please note that the payload format will slightly change in this new version. Each contribution will have a new "id" entry containing the filtering ID encoded as a big-endian bytestring (alongside the existing "bucket" and "value" entries).

We plan to fully launch the filtering ID feature (i.e. to Stable) later this year. We will notify this mailing list before launching.