What is the IDFA (Apple Identifier for Advertisers)?
The IDFA (GAID on Android) helps mobile marketers attribute ad spending: knowing that a new player launching your Game_A comes from an Ad that he saw and activated in Game_B.
This allows Ad providers to track a user across different mobile games owned by different companies.
Apple announced at the last GDC that, starting with the release of iOS 14 in a few months, app and game developers will need to ask users for permission to track them across apps and websites owned by other companies.
Why doesn't askblu.ai need to use the IDFA?
Every game has a specific way to be played depending on the type of game, the content, etc. That's why in askblu.ai, each game’s data is "encapsulated" in a "silo":
- each game has its own Machine Learning model
- no data from playing Game_A is ever used for Game_B, directly or indirectly (after processing)
- no data related to one game is shared with any other game on the platform
It is in the core design of askblu.ai to not track players across games, be it from the same or different publishers. That's why, from the beginning, we decided to not use the IDFA on iOS and the GAID on Android but random keys:
- a random key is generated when a new player starts a game
- the same player, starting another game, will generate another random key (so we do not know that it is the same player)
Which data does the askblu.ai SDK collect and for what usage?
Game developers will soon be required by Apple to provide information about what data the third-party SDKs collect, how it may be used and whether the data is used to track users.
The askblu.ai SDK collects only gameplay information:
- the sessions data (times and lengths of sessions)
- the starting of a “stage” ("level")
- the ending of a stage and how it ended (lose, win, quit...)
No personal data whatsoever is collected by the askblu.ai SDK:
- no data about the gender, age, etc
- no data about the player’s location
- no data about what the player is doing in other apps or games
The data collected is used to:
- feed a Machine Learning model (one per game)
- predict the risk of churn because of frustration or boredom
- display metrics and funnels (analytics)
This prediction is only used by the game itself (make difficulty easier or harder) and does not trigger any outside solution (push notifications, etc).