Learn in the textbookSearch Optimization, Effectiveness Measurement, and Delivery Technologies
DefinitionA/B testing (A/Bテスト) prepares two versions, shows each to a different group of users, and compares results such as clicks or purchases to pick the better one.
ExplanationChanging only a few elements makes it easier to tell which change caused the difference in results. Assigning users at random and running both versions in parallel over the same period keeps every condition other than the versions the same. The benefit is that the choice rests on numbers from real users' behavior rather than hunches or personal taste.
AnalogyIt is like handing out two flyer designs in turn, one per passerby, over the same period, counting which flyer brings more people into the shop, and printing that design from then on.
Memory hookThe name simply puts side by side the labels of the two versions being compared, A and B.