I have a puzzle game with N stages where each stage has 3 difficulty levels. Users have to finish current unlocked stage to access next stage, but they can access any difficulty level for unlocked stage.

As example, User A played stage 1 with difficulty level EASY, then he changed the difficulty level to HARD and finished stage 2.

I'm using Flurry Analytics to log as above: "stage1_Completed_Easy" and "stage2_Completed_Hard"

The problem is: there could be more users completed stage N than those who sompleted stage N-1 As example, there are:

  • 100 players completed stage1 EASY
  • 50 players completed stage1 HARD

  • 50 players completed stage2 EASY

  • 25 players completed stage2 HARD
  • 70 players completed stage2 NORMAL

I'm making assumption that each player only completed each stage once. This means from stage 1 to stage 2, I'm losing 5 players

I want to know which level and what difficulty that makes users stop playing my game

  • 2
    \$\begingroup\$ This is naturally that some players will stop playing just because they didn't like the game, and that is irrelevant to difficulty level. As for the level itself, a significant (compared to previous) user drop after certain level could indicate that players have difficulty completing it. \$\endgroup\$
    – pabdulin
    Commented Oct 14, 2014 at 13:36
  • \$\begingroup\$ While some players will leave because they don't like the game, those losses will be early, probably before level 10. The trick is to examine the rate of change and look for a spot where the dropoff spikes. \$\endgroup\$ Commented Jan 11, 2016 at 16:23

1 Answer 1


I actually do not know what level of liberty Flurry analytics give but you can, for example send and unique user_id with each event and with that remove people who played more than one level of difficulty or even count how many people retried what level. And with enough processing power you could generate a storyboard based on the events of you player, meaning: first they go to level 1 normal then half goes level 2 normal half goes level 2 hard.

Another solution is to send a counter for each player so you can do the same thing. That counter could increment each time you send a log to flurry. But once again that depends on your back-office

  • \$\begingroup\$ upvoted so the Community bot stops spamming this question up again and again. \$\endgroup\$
    – Philipp
    Commented Mar 9, 2016 at 9:27

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .