I am conducting an AB test on my site. I want to decide what is considered a good MDE (Minimal Detectable Effect) when it comes to click-through rate. I can understand you can choose MDE when your main success metric is revenue since you can just look at the cost of the experiment and calculate what the minimal change is to see an increase in ROI. Not sure how to determine this for click-through rate though. Anyone have experience with this?
There are tools to help you calculate this where you can define success however you like and it would test the significance of the result, but I would suggest trying to understand whether it is really meaningful to test click-through rate because if it doesn't actually match to an actual conversion then it could lead your research and testing down the wrong path.