Early behavioral indicators to an entertainment focused information system

Playing video games is a popular way to spend time and mobile gaming is one of the most growing entertainment industries in the world. Gaming is often associated with high level of motivation from the user as well as many negative and positive outcomes. Commitment towards games has invited countless of researches to examine what makes them so intriguing and motivating. This growing understanding gives developers more tools to design even better games and allows game-like features to be used in other contexts as well. This master’s thesis examines automatically gathered early log-data (n=100 000) from two free-to-play mobile games in order to create a model for retention. A model created using early log-data (first three days of play) creates opportunities to recognize potential players in an early phase and to evaluate early iterations of games that are in development. Furthermore, individual features are analyzed to study, what are the factors that influence coming back to the game at a later point (30 days after the installation of the game). The research questions in this thesis are: 1) Can commitment towards mobile games be modeled using early log-data? 2) How accurate predictions the created model can do? 3) What are the most important in-game features that predict retention? The model is created using a decision tree analysis, which was selected as a method due to its transparency and because it has been used before in earlier studies with similar designs. In both games, the rate of coming back to the game after 30 days of installation was 7.6%. A working model for retention was formulated which was able to predict coming back to the game with 33% accuracy. The most important in-game features that affect retention were the number of victories, the number of starts and the number of in-app-purchases during the three-day period after the game’s installation. Surprisingly, in-game rewards and achievements were the most insignificant features when predicting retention although they are often specifically designed to elevate user motivation. These results can influence design decisions made in game development by setting the focus on the factors that influence player commitment and behavior. Achievements and in-game rewards might feel too artificial and superficial compared to winning in game. If the system gives direct feedback of the effect of time and monetary sacrifice to the player’s performance, one might be able to reduce the number of players that decide to leave the game. The results also can be used to examine how game-like features are used in non-game systems where the goal is to tie together the high-level of motivation seen in games and socially impactful endeavors. Further studies of in-game behavior might also give new insights on game addiction and its negative effects on player well-being and business