Tell me if there is something wrong with my reasoning below:
There is percent rollout set to 50%. I guess that means only half of the players will be affected by the experiment.
But then there is also a control variant (required). Is that not the same thing? So, half of all the players will enter the experiment based on Percent Rollout, and then just 75% of them will actually test a new thing, while remaining 25% will be unaffected, as if they didn’t enter the experiment at all? (So, 37.5% of all players will actually test a new thing, and not 75%)
I think I should put the percent rollout at 100%, and just change variable percentages. Because now, the remaining 50% that seemingly wasn’t in the experiment, are de facto experiencing the Control version of the game BUT their data (playtime, retention etc) is not recorded (lost).
Is there any reason not to put percent rollout at 100%?