One way for auditing selection bias issues is running what's called an A/A test, like we saw earlier. So, if you actually run an experiment where there is no difference between the treatment and control, you shouldn't see a difference in the end result. There should not be any sort of change in behavior when you're comparing those two things.

An A/A test can be a good way of testing your A/B framework itself and making sure there's no inherent bias or other problems, for example, session leakage and whatnot, that you need to address.