The only thing that upsets me about this article is the part where he cites an article saying that most people don't understand statistical significance in A/B testing as a justification for not using it when in actual fact he has the exact same statistical issues by not using it and suddenly loses a bunch of control variables.