Hello, Just wondering if anyone has come across this issue before. We had a test running with a smaller traffic allocation. but the results strongly and significantly favoured the Variant. As it was only a portion of paid traffic being sent through the test, we upped the numbers almost triple. The week after, the daily results flipped. The variant was now losing. A month or so on and it is still losing. I cannot work out why increasing the traffic would suddenly flip it on its head.

how much traffic are we talking about?

first version was only getting 20-40 users a day each. To original and variant. after they increased, it went up even more than I thought. 200-250 a day. each.

how long did you run it?

Been running since Feb.