A/B Testing uses data to challenge player assumptions and to test new ideas or mechanics in the game. Good A/B testing can lead to increased engagement and/or monetization. Traditionally A/B testing has a control version of a game a version that the developers want to try out to see if it increases purchases, enjoyment or engagement. For example if you have a split population of testers, one plays the ‘control’ version of the game. Other groups play two variants of the game, you measure the response (this group played for longer/spent more money/played 100% evil) and then choose a ‘winning’ model. A/B testing is used mostly in games with in-game purchases, (Free to play etc.) or monetization models and examines the best approaches towards getting players to make purchases. The A/B model can be used for almost any feedback, to see if players finish any tutorials, if they watch or skip cut-scenes or if they ‘buy-in’ to the world.
