Replicating Google Analytics' user_engagement metric in Amplitude

Hey @Anthony 

If it’s just the DAU metric in true sense, then a simple event seg chart using your “Active Event” with Uniques against a daily time frame will give your DAU metric. e.g.https://analytics.amplitude.com/demo/chart/new/77bpa14

Looks like the definition Google uses for [GA4] User engagement  in there relies on the time the app screen was in the foreground.

If it’s more on the session side, then the User Sessions chart will be a better choice. I would suggest taking a look at how Amplitude defines sessions both on mobile and browser  - https://help.amplitude.com/hc/en-us/articles/115002323627#how-amplitude-defines-sessions - so that you can try mapping your requirements here accordingly.

Let me know if I have interpreted your question correctly. Hope this helps.

We’re using the Amplitude Android SDK and trying to trigger the right events to be able to match Google Analytics. We are including events like “view homepage” and “view splash screen” and “view loading page” but when I include the view loading event, our number is higher than Google and when we exclude the view loading screen event, our number is lower then Google. 

As you can see, we’re missing the Google target by about 10k users (about 10%) in both directions.  

This has been one of the frequently discussed topic in the community when users are trying to match the numbers from their Google Analytics and Amplitude implementation.

Here’s an article which can help you understand some of the data discrepancies - https://help.amplitude.com/hc/en-us/articles/115002383247-Understanding-the-causes-of-data-discrepancies-between-Amplitude-and-third-party-platforms

Thanks! In our Native Android app today, it’s sometimes hard to tell if the events were triggered while the application was open and in the foreground vs triggered/emitted from background processes. Maybe we can look into adding an event property like “application_in_foreground” (true/false) so that only events that were emitted while the app is in the foreground would be counted toward our DAUs. Does that sound practical and useful or spark any alternative suggestions?

Yea, if you want the user activity to be counted only when certain type of events are triggered ( i.e. foreground ), then this approach could work.
Or you can also mark your background process events as inactive in your schema and use the Any Active Event as part of your DAU calculation instead as Any Event.

Thanks!