Player types in a mobile game

Abstract

The aim of this thesis is to find out what types of players are most prevalent in a popular mobile game created, published and maintained by the case company. The case company wishes to understand their users and customers better by exploring what kind of player types play their game, what motivates these players and what kind of in-game content is most engaging and enjoyable for each player type. As the game has millions of players, it is necessary to build an analytical model that can be used to determine player types by estimating how different measurable player actions tracked by the game determine the player type. The model built in this thesis will place values on each significant measured action found in the data to move a player along two axes, with the end result determining the players location on the player taxonomy graph and thus their player type.

This study uses mixed methods approach to determine the player types. It consists of a literature review on the importance of understanding customers, player taxonomies and mobile games. The data used is sourced from a vast database collected and maintained by the case company. The database contains extremely detailed numerical data depicting the actions taken by the players within the game. In the first part of the analysis, the most suitable player taxonomy attributes determined in the literature review were coded with the appropriate columns of the database consisting of the player data. The combined values for each individual player type dimension were then calculated into a new table, portraying the players type in four dimensions: killers, achievers, socializers and explorers. Using the values in these dimensions the player is then placed into its position on the player taxonomy graph, which in result determines the players overall player type.

The most significant findings of the study indicate that from the 134 561 players analysed, the vast majority showed up as explorers (130 813) with the remaining 3748 being achievers when placed on the original player taxonomy graph introduced by Bartle (1996). Using weighted values that attempt control or the number of attributes for each type helps other player types show up. Killers (167) and socializers (1608) show up but achievers (32504) and explorers (100282) are still the vast majority.

The analysis shows that the player type distribution created through data analytics is subject to the possibilities provided by the game. If equal opportunities are not provided for all player types to be present in the data, equal player type distribution cannot be achieved without weighing the data; instead, player types distribution will be skewed towards the types that are represented in more values. Findings from the game data also indicate that player types that relate to interacting have a distinctly different distribution on the player type graph compared to players that act or interact with the game.