Hey, we're running an experiment at the moment with a 50/50 split. Some colleagues in the office who saw the new UI initially are now reporting that it has reverted for them. Just trying to understand how that could happen?

While generally for live experiments on production builds, once a user is allocated to a variation in an experiment, that user would remain in that variation for the duration of the experiment until you choose a winner, there are ways that they could receive a different variation or be simply excluded from it.

One scenario is that if you change the distribution of an experiment to essentially expand or restrict the scope of the experiment, this would entail removing filters or adding filters.

The other scenario would be for internal devices. If you’re using dev builds with different bundle identifiers, then Taplytics could possibly recognize it as a different device; thus, the device receives a different variation.