Solved

Fraction of users that perform an event each day of the week

  • 19 January 2022
  • 4 replies
  • 103 views

Badge

What’s best practice for finding the fraction of users (who perform event A) that perform an event A (at least) once every day of the week?


I have tried a solution ​​​​​​to use a Segmentation chart with two segments:

  1. Users who performed event A >= 7 times
    any time in each week
  2. Users who performed event A >= 1 time
    any time in each week

On the left hand side I chose Any Active Event,
using this formula: %: UNIQUES(A1)/UNIQUES(A2),
with frequency weekly.

My understanding is that this will give me the fraction of event a users that performed event A >= 7 times of the users who have performed event A >= 1 time. However, the “in each” option will include user who performed event A at least X times during each chart interval, which does not met my intended wish to guarantee that each user in segment 1 have performed event A at least once each day, for example a user would also be included if they performed the event 5 times on Monday and 2 times on Saturday...

In short, is there a way in Amplitude which can only count user that have performed an event at least once each day on a weekly basis? Or do I simply need to analyzing the raw event data outside of Amplitude?

Grateful for any help, explanation!

Thanks in advance,
Nils
 

icon

Best answer by Saish Redkar 19 January 2022, 17:09

View original

4 replies

Badge

Update: I have tried a Stickiness chart which seems to solve my issue of only including users who perform an event at least once on the exact number of days.

Although, any feedback/confirmation on this is still greatly appreciated. 

/Nils

Userlevel 7
Badge +4

Hey @Nils 
Using the Stickiness chart is the best way currently to handle your use case. You can toggle between the Cumulative and Non-Cumulative option to analyze your chart accordingly.

Badge

Once again thank you @Saish Redkar !

Userlevel 5
Badge +4

Hi @Nils !

In addition to the Stickiness chart, you can probably use the in-line cohort offset to help you define a segment of users who have performed event A at least once each day. You can read more about offset here: https://help.amplitude.com/hc/en-us/articles/360040965352

Here’s an example of what I was referring to: https://analytics.amplitude.com/demo/chart/new/554k6f0 Segment 1 is looking at users who did Event A today, 1 day before today, 2 days ago, 3 days ago etc. 

For the offset to work though, you will need to use the daily interval. 

So if you want the chart to be “weekly” calculation, you could consider using RollingWindow: https://analytics.amplitude.com/demo/chart/new/m1bavof. Note that RollWin is on a rolling 7 day rather than calendar weeks

Hope this helps!

Reply