Conversion fees go up, however, you can??t pinpoint what adjust was responsible. If you take a look at one variable at any given time, you could isolate the effects of every alter and obtain much more accurate success. It doesn't matter how persistently your A/B test fails, its eventual good