Solved

Successfully Using Application Installed in Funnels

  • 9 September 2022
  • 2 replies
  • 232 views

Hi - frequent listener 1st time caller! I’m hoping for some guidance around how to use Application Install in my funnel analysis. I’ve broken out my questions below:

  1. I don’t see ‘Application Installed’ in any of Amplitudes funnel literature, so is that my 1st mistake?
  2. In the past I was only able to use ‘Application Installed’ in an event segmentation chart - as soon as I used it in a funnel I showed 0 conversion. We figured out that this was due to our Segment integration and updated that about two weeks ago. 
  3. This has led to some unanticipated consequences (doesn’t it always). Our ‘Application Install’ numbers shot up. It appears that we are now tracking ALL installs - both net new and re-installs. 
  4. When I filer the Application Install number by new users it drops it down but over the last 2 weeks its has been 92% of the number that I see when I look directly at my installs in the app stores. Is that a discrepancy that I’m just going to have to live with? Is there anything else that I should be aware of around filtering by new?
  5. My engineering lead is pushing me to use the ‘New User’ event instead that is more closely aligned with my store installs but I don’t really understand what that means as it includes ‘inactive events’.
  6. My ‘New User’ event has also plummeted, I assume because we are now successfully linking anonymous installs with the rest of that users journey. 

Top level: should I be using Application Install in my funnel, and is it wrong to filter it by new users? Is ‘New User’ and adequate proxy for installs and can I use that instead?

Graph that displays the changes since our push. 

Thanks so much for any guidance that you can provide!

 

 

icon

Best answer by Denis Holmes 12 September 2022, 13:06

View original

2 replies

Badge +10

Hi! I’ve seen most of these things happen before so I feel your pain! I’ll try to cover a few of your points.

Using New User is generally a good approach. It is defined as the time period in which Amplitude first receives tracking events for that user. Including inactive events is not usually a problem, because for any Amplitude event to be firing it must mean your app has been installed. One thing to be aware of is if you had users before you implemented Amplitude and they come back, they will appear as “New” the first time they come back. This can often rule out using “New User”.

If you can’t use “New User”, I recommend instead you pick an event early in your UX that would only apply to New Installs, not app updates (e.g. how early is your user signup?).

 

Userlevel 6
Badge +8

Hi @blasky !
 

Excited to see you writing in! I like Timothy’s answer above but want to add my own information to it too!

 

  1. I don’t see ‘Application Installed’ in any of Amplitudes funnel literature, so is that my 1st mistake?
    Is the Application Installed in your own organization but not in Ampliutude’s, correct? This would just be an event that you have ingested by that name. It is an event that has been ingested called “Application Installed”. If you are still unsure, do let me know where you see it and I can help explain. 
  2. In the past I was only able to use ‘Application Installed’ in an event segmentation chart - as soon as I used it in a funnel I showed 0 conversion. We figured out that this was due to our Segment integration and updated that about two weeks ago.
    This could be that no one is completing Application Installed in a certain funnel. Can you link the two charts here? I can then better outline what is happening. 
  3. This has led to some unanticipated consequences (doesn’t it always). Our ‘Application Install’ numbers shot up. It appears that we are now tracking ALL installs - both net new and re-installs.
    This would sound like an implementation issue of sorts. Can you show me a chart?
  4. When I file the Application Install number by new users it drops it down but over the last 2 weeks its has been 92% of the number that I see when I look directly at my installs in the app stores. Is that a discrepancy that I’m just going to have to live with? Is there anything else that I should be aware of around filtering by new?
    When looking at New Users, you are looking at that event if it was done in the first calendar day of use of a user in Amplitude. Like Timothy explained, New User only counts events for a user’s first calendar day of being recorded in Amplitude. So if the event is not on the same day as the user’s first use, it will not appear in a chart when looking at new users. 
  5. My engineering lead is pushing me to use the ‘New User’ event instead that is more closely aligned with my store installs but I don’t really understand what that means as it includes ‘inactive events’.
    Inactive events are ones you have marked as inactive in your Govern. This might be something like receiving an email or a push notification. 
  6. My ‘New User’ event has also plummeted, I assume because we are now successfully linking anonymous installs with the rest of that users journey
    Correct, users are probably starting to merge under their correct profiles. There may also just be less new users coming to the app. 

    I hope this helps for now!

    Kind Regards,
    Denis

Reply