Events Lost Using Google Analytics for iOS SDK 2.0b3

Question

I recently implemented the Google Analytics iOS SDK v2 in the number of iOS apps , and I noticed that there are some seemingly big discrepancies in the generated reports. My main concern is that the total number of events for my tracking ID is about 50% lower than my internal metrics show what it should be. I spent some time trying to get rid of possible problems.

At first I thought that maybe I didn’t make Google tracking requests from the app, but they were all verified. Then I tested the theory that metrics did not exit the local database over the network. After setting up wirehark to monitor network traffic, although it also seemed to work correctly. Then I tried to process events more slowly, as I thought it might be the fact that events are fired at short intervals, which could be a problem. After slowing down the pace of the event no more than 1 per second, and then to a maximum of 1 time in 3 seconds, I still observed a loss of about 50% when monitoring my test sessions in the web interface. In addition, it seemed somewhat random which events would be deleted in the reports that I generated.

Here is another information about my analytics configuration:

• 60 second send interval
• Manually Starting Sessions
• Using multiple trackers

It is strange that the Android versions of the same application do not have the same problem.

If anyone had similar experience or have any suggestions, they would be greatly appreciated.

Answer 1

After contacting the Google Analytics team, they confirmed that beta versions 3 and 4 of the iOS SDK have an error that is responsible for the understated report that I see.

Answer 2

There is information in https://developers.google.com/analytics/devguides/collection/ios/v2/events (typo, Android → Analytics):

Note. The Google Android SDK for iOS can throttle events, as well as other hits, if a large number of calls are made to send in a short period of time.

So, I think your problem might be related to this. If you track traffic using wirehark and your application sends the proper calls, I think you cannot do this. Perhaps try to increase the interval between calls to a truly undeniable value - for example, 30 seconds - and then calculate the difference.