How to Retain Players through Dynamic Quality Adjustment in Video Games

Subscription model is one of the two commonly-used business models in the video game industry. It has been applied by some famous games such as the World of Warcraft, and EVE. Thus, how to retain players and make them willing to renew their contracts is a critical and long-running quest for game operators. Relying on characteristics of video games, we present a stylized dynamic programming model to help video game operators decide what quality level they should provide in different periods to retain players. With the assumption that both players' memory of past service experience and network externalities can affect the resubscription decision-making process of players under the condition of profit maximization, we use a Q-learning algorithm to simulate the decision-making process off-line. We show that the game quality levels will converge to steady state no matter what the initial game quality is. However, the convergence path would be longer when the initial game quality is low. Then, we demonstrate that the trend of perceived quality indicates a "high-low-high" pattern when players overestimate initial game quality. The results also reveal that if network externality intensity is high, it is negative for game operators to lift game quality.