How Not To Run An A/B Test
Just because Google Optimizer tells you that your new version has a high percentage chance to beat the original doesn’t necessarily mean that the new version will convert better.
While this sounds completely counter-intuitive, you need to have enough visitors to be sure that you are running enough traffic to each version page to ensure that statistical errors aren’t skewing your data.
This article will show you how to ensure that your experiments are giving you the right data, you don’t want to declare the page that converts lower the winner!
By Evan Miller