7

I'm close to opening up access to a web application that I've built, and I wanted to make sure that I can link updates to the app to errors, or other changes in analytics data. My first thought was to put the current git ref in a custom variable that is tracked with every event. But within analytics systems, you don't have access to the git log, so all you have is an opaque ref that you have to cross reference with your application's repository. I could also put tags in the repository with release numbers, but all I have then is a more coarse-grained identifier, that I still have to cross-reference somehow.

So I'd like to ask the advice of people running web applications in production: what has been most useful to you to know what change caused a certain change in user behavior or application performance?

bigblind
  • 1,415
  • 1
  • 13
  • 17

1 Answers1

4

I add the version number as a custom dimension in Google Analytics. Providing your release cycles are not too long, this gives a good indication of how each version impacts usage of your application. As each release generally contains one or two main new features (along with tweaks and bug fixes), there's enough granularity for cross referencing.

Other things you might want to consider as custom dimensions:

  • Which plan a user has subscribed to
  • Which specific month or year a user signed up
  • From where did you acquire a user

Google Analytics Custom Dimensions and Metrics: https://developers.google.com/analytics/devguides/collection/analyticsjs/custom-dims-mets

kerrin
  • 290
  • 2
  • 9