How Mixpanel helped me build my VR puzzle game

  • 14 January 2021
  • 1 reply
  • 162 views
How Mixpanel helped me build my VR puzzle game
Userlevel 1
  1. In a few sentences, can you describe how your company and/or teams use Mixpanel? Which teams/roles use Mixpanel? What features do you use most? What are the main KPIs your team tracks?
    I recently released a VR puzzle game called Cubism for Oculus Quest and PC VR headsets, in which players assemble increasingly complex geometric shapes out of colorful blocks. I developed this game on my own in my spare time over the course of 3 years. The game uses Mixpanel's Unity integration to anonymously keep track of certain events in the game. During development, Mixpanel helped me evaluate puzzle difficulty by aggregating anonymous playtest data, and now that the game has released, it's helping me keep track of usage statistics on each platform.
     
  2. What challenges are you trying to solve using Mixpanel? What goals are you trying to achieve?
    Puzzles in the game are organized by difficulty, but accurately estimating the difficulty of a puzzle when designing it turned out to be really difficult with traditional playtesting, since people's skill with Cubism's puzzles can vary greatly.
    I ended up addressing this problem by testing puzzle designs with larger groups of people via a demo of the game published to SideQuest (an unofficial store page for the Oculus Quest), and using Mixpanel to aggregate anonymous statistics on things like solve percentage, average solve time and moves per minute for each puzzle. These statistics gave me a better basis to evaluate difficulty and compare puzzles, allowing me to improve the difficulty curve of the game and improve the overall experience.
     
  3. Has any discovery you made on Mixpanel surprised you in any way? Why?
    Now that the game has launched, and many more people have played the puzzles then during development, it's been very interesting to look at the same statistics I was tracking during development (average solve percentage, solve times, etc.), and seeing where I got the difficulty curve wrong now that the data set is larger and more accurate. It's clear to see which puzzles on average people are struggling with, and it's helping me understand a bit better what exactly makes for an easy or a difficult puzzle.

1 reply

Userlevel 6
Badge +3

Love this story! Thanks for sharing @Vanbo 

Reply


Mixpanel