University of Michigan researchers create better matchmaking algorithms for multiplayer games

University of Michigan researchers studied the matchmaking system in a popular first person shooter, and have found that players with different levels of engagement respond to different incentives to continue playing. The research will help game developers stand out in a crowded market, by fine-tuning the matchmaking systems according to the players, instead of randomly putting a bunch of people together in a game.

The researchers categorised the players into three levels of engagement, low, medium and high. It was found that the most skilled players, continue to play the game for achieving high scores and improved rankings, that continues their domination over lesser skilled players. The most skilled players are not interested in being challenged, and are comfortable enough in their winning streaks, and are more interested in achieving victories.

At the lowest level of engagement, both rankings and challenges have a modest affect on retention. At the middle level, which is the most populated level in any game, the players respond strongly to both being challenged, and bagging achievements. Most players in the middle level are interested in improving their games, as well as climbing up the ranks. Puneet Manchanda, professor of marketing, said “That was the surprise and it was hard to articulate before we saw the data. They want to play to better themselves, not just to score a higher rank.”

The researchers have developed a matchmaking system that matches players based on their engagement to the game and level of play, achieving a four percent greater customer retention of 7.8 percent as against the 3.8 percent offered by random matchmaking. The algorithm is fast, scaleable and works in real time.