GameAnalytics

For issues or assistance, please click here to contact Support
Unreal integration makes no sense
I just spent an afternoon trying to get the Unreal integration to work. To be clear, I'm installing from GitHub (which already had issues because if you try to download the Source.zip from releases, the big .lib file isn't there, instead you get the LFS stub). I'm using UE5.4.4 and C++ (not Blueprints). I almost gave up, but I dug into the issues and here's what I found. The instructions from https://docs.gameanalytics.com/integrations/sdk/unreal seem to indicate that GameAnalytics acts as an analytics provide for Unreal. This sounds great, except that Unreal doesn't run analytics provider in non-editor builds - so game/cooked builds wouldn't work properly. This is obvious if you look at the function "void FEngineAnalytics::Initialize()" in "EngineAnalytics.cpp" (this is Unreal's code). As a result, simply calling UGameAnalytics::initialize as the instructions indicate is insufficient, and the analytics engine is misconfigured. Besides, all the configuration that exists in the Project Settings is completely ignored, because it's only accessed when a session is started using the analytics provider interface. It was also very inconvenient that the SDK-CPP lib came in binary format, preventing me from debugging it and figuring out the issue much faster. I even had to modify the integration to add a log handler to understand why some events (resource) were not being sent correctly. Overall, I'm disappointed by the integration process. I don't know if going with Blueprints would've been better - though it that still relies on the analytics provider interface I doubt it. My advice would've been to essentially expose a very thin interface over the CPP lib, just transforming from FString to char*, and then make those available to Blueprints as static functions. The module already handles Startup and Shutdown, so integration could be seamless.
2
·

under review

About Data Reliability
Hi, there! Thanks to GameAnalytics, I can easily operate the project and check the chart. There were some things I wanted to improve in displaying the data, but they have been resolved well. This time, I would like to talk about the most important issue. That is, data reliability. Charts, that is, data, are the most important elements. However, in the case of the New User data, there are many cases where seriously incorrect data appears. If you check the provided screenshot, you can see that the number of New Users suddenly increases sharply during a certain period, while Retention decreases sharply. We can be sure that the sharp decrease in Retention is due to incorrect counting of new users. When you go to the store and check because you are excited about the sudden increase in users, you will find that the actual data is not like that. This phenomenon does not occur only in certain countries. It occurs in almost all countries where we distribute games, including the United States, Japan, and Taiwan. I am well aware that it is difficult to create completely clean data. There are various methods for measuring new users, their pros and cons, and various difficulties. This problem even occurs similarly in Google's Firebase. However, this happens much more often with GameAnalytics. Another service we use, Adjust, has very high data reliability. As a result, we do not use some data from GameAnalytics, such as New User, in our decision-making. I am curious as to why this happens and how we can improve it. Is it because of testing in the development environment? Or because we use test devices? Or because a large number of users suddenly delete and reinstall the app at once, or use programs like App Player?
0
Load More