User buckets, or buckets for short, describe user attributes that result from a random equal distribution of users in segments. Their allocation results in several identical target groups that differ only in the value of the bucket.
The value or number of buckets is not firmly defined and varies depending on the tool used. For example, some tools allow a value division of 1 – 100 or more. User buckets are often used in A/B- or multivariate tests in order to compare the results of several variants with the same target groups.
Example:
Three different variants (A,B and C) of a layer are to be tested on identical target groups. Additionally, a control group is to be added.
Assuming an equal distribution in 16 buckets, buckets 1-4 fall into segment A (layer variant A), buckets 5-8 fall into segment B (layer variant B), buckets 9-12 fall into segment C (layer variant C) and buckets 13-16 fall into segment D. This last segment is defined as the control group.
In a comparative test, such as the multivariate test, it is then possible to determine which campaign is most effective. In addition, a possible uplift can be identified by the control group.
In contrast to a percentage distribution, in which the user moves from one segment to another, users remain in a bucket once they first fall into it. From this point on, the user will always see the same campaign over and over again, unless the distribution of buckets is redefined. At the same time, users in the control group will never be addressed by a campaign.