Randomization

the process of randomly assigning inputs to different versions in an A/B test to distribute confounding variables evenly across groups so that any systematic difference in outcomes can be attributed to the treatment rather than background conditions