7 min read
5 ways your A/B test is lying to you (and the engineering fixes)
A/B test results mislead more often than teams admit, and almost always because of how the experiment was instrumented, not the statistics. Here are the five failures I see most, and how to build so they don't happen.
Read the essay