> You're seeing a difference between Control and Variation in an A/A test is because a very small number of visitors have been tested.

Well, we don't actually know how many samples the OP has, do we?