attribution-reporting-api
attribution-reporting-api copied to clipboard
Daily data for campaign optimization
I am following up the presentation of 09/21 meeting.
Getting the report fast enough (feeding the algorithm with at least daily data to react and invest in new “contexts” that drive conversions) is very important for campaign optimization.
In the context of Post View "event reporting" we would have the data at the end of the expiration window (possibly 30 days) which would be extremely harmful for campaign optimization.
How do you think we can solve this problem ? May be adding another reporting window and making sure we still guarantee user privacy ?
How do you think we can solve this problem ? May be adding another reporting window and making sure we still guarantee user privacy ?
I see two ways of solving this:
- Add another reporting window at the cost of some privacy (or giving up some other feature)
- Perform user subsampling where fractions of users have artificially small expirations with small delays, and other users have normal lookbacks and longer delays
Obviously I think (1) will be higher utility, but I am interested in how (2) would perform given that it has no privacy loss.
Those two solutions seem very interesting.
1.Add another reporting window at the cost of some privacy (or giving up some other feature)
Seems pretty straightforward indeed.
2.Perform user subsampling where fractions of users have artificially small expirations with small delays, and other users have normal lookbacks and longer delays
It seems interesting to get a part of the data beforehand (with the subsampling "ratio" of course to be able to estimate a number of 'events').