Our analytics team is interested in having a "holdout group", for example when bucketing users a certain percentage of traffic will not enter any tests. Obviously there will be a small percentage of traffic this is true for, but I think their goal is to have this be a larger percentage than what it would be naturally.
We can imagine maybe writing a proctor test that buckets traffic this way and then having a requirement for every other test be that the user is not in the "holdout group". However, it seems like a pain to have to write that for every test.
What I sort of want is to have a nested test approach, where only if the user is not randomly bucketed into the holdout group, then they are exposed to a test.
Is there support for this in proctor that I'm not seeing?