How Long Will Your Experiment Take to Run?

If you're trying to improve registration rate and that's 50%, then your baseline conversion would be 50%. The second variable is minimum detectable effect. So we want to see at least a 0.5, we only care about differences greater than 0.5%. And that's kind of the basis of our experiment. Now, if your minimum detectable effect is smaller, your sample size will end up being larger,. If you want to detect a 0.01% change, you're going to need a huge sample size. But if you want to detects a 5% change, it's going to be much smaller.