Experiment versions
In the Proba.ai service it is possible to edit experiments using classical user distribution. This is necessary if one of the variables of the experiment begins to show insufficient results from the very beginning.
The service provides the ability to change the amount of experiment variables, as well as traffic redistribution. When making changes to an experiment, it is incorrect to compare the sample before and after the changes had been made. Therefore, the experimental version has been implemented in the service. The new version is a log of changes occurred during the experiment.

A new version of the experiment is created if, when editing:
  • the quantity of variables changes;
  • the distribution of shares between the variables has changed.

A new version is not created:
  • When the experiment title or description is edited;
  • If something has changed in the options, but the participants in the experiment have not yet appeared (the app is still under moderation).