Bucketing - Featurevisor

Bucketing is the process how we assign users to a specific variation of a feature.

Factors

Few factors are involved in the bucketing process:

Feature: the feature key (name of the feature), and the weight of the variations
Attribute: the bucketBy property of the feature (usually userId)
Rule: the rollout rules, and their percentage values

Bucketing process

When a user is evaluated for a feature, the following steps take place:

Create a bucketing key from the bucketBy attribute of the feature, and the feature's own key (the name of the feature)
Generate a hash from the bucketing key, that ranges from 0 to 100
Iterate through the rollout rules, and check if the segment has matched and the hash is within the range of the rule's percentage value
If a match is found, the feature is meant to be exposed to the user
Then find the variation that the user is assigned to, based on the weight of the variations and the hash

Consistent bucketing

The bucketing process is consistent, which means that the same user will always be assigned to the same variation of the feature. This is achieved by using the same bucketing key for the same user, and the same feature.

It is possible to maintain consistent bucketing as long as the feature's rollout rules' percentage keeps increasing over time. The expectation is we will always gradually increase the rollout percentage of a feature, and never decrease it.

If the percentage of a rollout rule decreases, then the bucketing process will not be consistent for all users any more. Even though Featurevisor tries its best to maintain consistent bucketing, it is not possible to guarantee it in all cases if the percentage value decreases.