Putting Machine Learning at work for mobile games
GameTune versus askblu.ai
NOTE from Unity on the GameTune web page:
"We created GameTune with the intention of making Unity’s machine learning
approachable for all studios. However at this time, we have decided to pause
further development, and discontinue the GameTune service.
We will keep you informed as we have new updates to share."
Below is a quick comparison between GameTune and askblu.ai.
Our information about GameTune is public information, available to anybody on the Unity web site and their SDK documentation.
You can also read our article on MEDIUM: "Machine learning for mobile games: GameTune versus askblu.ai"
What is the purpose as described by the provider?
What metric does the machine learning aim to improve?
Is the solution design based on a game theory?
How do I create a request for personalization?
Do I get any feedback on the values I want to personalize?
What user-data events do I need to add in my game?
How can I measure the efficiency of the solution?
Will my own game's data be used for other games?
Do I need specific staff with data competencies to use properly the service?
What is the purpose of the service as described by the provider?
Which metrics does the service aim to improve?
Is the service designed around a gamer psychology fondation?
How do I create a personalization request?
How Machine Learning models are created and used by the service?
What user-data events do I need to add?
How can I measure the efficiency of the service?
Will my games's player data be used for other games?
Do I need a specific staff with data competencies to use the service?
"Adjust your game in real-time to deliver the optimal experience to your players."
Retention, Revenu, Conversion, Other (to be defined by studio)
Set names and parameters, both in the SDK and in the web dashboard.
One model trained with data coming from many different games.
You have to find which data could be useful, and to implement the events, in the SDK and on the web portal.
You have to set yourself some A/B metrics with an external analytics tool.
Yes, the service is driven by your game's data but is also leveraging other Unity game's data.
Highly recommanded (define reward, requests and player events, set A/B metrics...)
(Happy Blue Fish)
"Keep more players engaged with real-time in-game personalization."
The service is optimized to improve player retention annd LTV
Yes, the Cognitive flow theory.
Use the pre-defined requests, in the SDK only (no web portal setup).
Each game has it's own Machine Learning model, trained only with the data from its own players.
Just set the 2 pre-defined events (stageStarted and stageEnded), only in the SDK (no web portal setup).
All metrics and funnels included in the web portal display data for players with and without askblu.ai personalization.
Never; And we do not use the Apple IDFA. And we use no personal information, only behaviour data.
No, the service is ready for your team of producers, game designers and developers.