I'm intrigued by post-game screens. They display many interesting tidbits of information: stats, level progression, match highlights, and often automate the player traveling to a consecutive match. However, many players quit before automatically playing the next match. In what ways can we change the post-game screen to keep players more engaged?
For our upcoming title Kabounce, a multiplayer pinball game in which you control the ball, I’m doing research into this post-game state to maximize the flow of our core game loop.
Popular games such as Rocket League, and Overwatch, share similarities in their post-game screens. To give an example of typical content, below are the elements used in Overwatch:
OVERWATCH POST-GAME USER INTERFACE SCREEN (STEP 1)
OVERWATCH POST-GAME USER INTERFACE SCREEN (STEP 2)
In comparison, we can see many commonalities and other small differences in the design, stats, and layout of Rocket League:
ROCKET LEAGUE POST-GAME USER INTERFACE SCREEN (STEP 1)
ROCKET LEAGUE POST-GAME USER INTERFACE SCREEN (STEP 2)
By comparing these, we can begin to design our own ideas about what might make a successful post-game screen, but we need to do tests to understand more about their impact on player behaviour and why they leave games at this point.
Initial hypothesis: Players tend to leave early because the timer is too long.
To confirm my suspicion, I asked 15 players with various player profiles to answer a questionnaire. One of my questions included multiple pre-determined selections of options that was the same for each game. While Kabounce tests results at this stage were unreliable - all players have some form of relationship to the development team. However, personally I still found it interesting to share and look at these early results.
The most noticeable results were:
These results suggested, along with corroborating questions, that players are likely to want to play another game; players who played Overwatch most likely enjoyed the game; the time of the end game screen has some correlation to the desire to not wait, and possibly exit the match.
However, researching other studios’ games limits some of the information we start with, which can have the following risk factors influencing the results of any data gathered:
To apply these to Overwatch, our results might change due the availability of the game. All players either downloaded it during a free weekend or they purchased it directly. These will give a different sense of investment in the game before it is played.
Rocket League and Kabounce both support split screen, meaning that not all questioned players will have downloaded or purchased the game themselves, which creates a very different group of players compared to Overwatch. Some players who only played less than an hour in split screen with a friend did not select that they ‘wanted to play another match’, so it is possible that these players in our study are not the correct target audience of Rocket League or Kabounce.
All players who tested Kabounce also had the chance to do so for free, potentially influencing the test results. Because of this, I formulated a new hypothesis, focusing on one game instead:
Adjusted hypothesis: Overwatch players tend to leave because they want to play another match more quickly, and the Overwatch post-game screen does not provide a quick opportunity to skip the timer.
As additional research to confirm my hypothesis I played 16 online matches of Overwatch, measuring the activity of 152 players. I recorded the post-game screen flow after each match, and documented unique players; confirmed if they left or stayed throughout the entire duration of the post-match screen. In these play sessions I only observed 1 player pressing the vote button to ‘stay with current team’.
Most of these players I observed played hundreds of hours. We can consider that the total amount of playtime, length of current play session, and external factors probably influenced these results; they were not conducted in a ‘sterile’ test environment, or with access to all variables. However, initial observations do indicate a high player exit ratio.
For us to analyse data in a more controlled environment, we want to log the playing behavior of players anonymously in Kabounce, i.e. logging time spent at various screen stages in the user experience over the entire length of their play session, and recording the length of their play session. We will also need to log their total played hours and their matchmaking rank to ensure we can categorise players in roughly equal groups.
By performing ABC testing we may be able to determine which of the following methods, if any, have the biggest impact on player exit ratios between consecutive game sessions:
Later this year I will be able to share the results of my study. We have three options to try, two of them having a direct impact on the post-game menu flow, and the third only adjusting player perception. Which will be most effective in getting players to stay in Kabounce? We’ll let you know!
Thanks for reading, and you can see the Kabounce trailer below:
Tim Baijens is a co-founder of Stitch Heads. He is a professional developer and also researching game design as part of a Masters study at NHTV Breda University of Applied Sciences.