A Split Test is a method of conducting controlled experiments with the goal of measuring which of two approaches has the best results. You divide your audience into two similar halves and show them two different versions of the same thing and then measure which one performs best. Like twins, they are almost the same, but not quite. The first, original version (A) is called the “control” and the new, alternative version (B) is called the “challenger”.
(This practice is under development and is part of the book Startup, Scaleup, Screwup.)