In many of the FME SDKs out there, there is a concept of a bucketingKey for ensuring splits evaluate to proper treatments based on hashed keys and percentage-based distributions. Ideally, this should be available in OpenFeature SDKs.

A possible implementation would be to regard a certain kvp in the evaluation context to be the bucketingKey. Currently, this same principle is applied to a targetingKey; if a kvp in an evaluation context has key targetingKey, then the the provider implementation picks that value and applies to the Split native (user) key. The same principle can be applied to a kvp with key bucketingKey.