Solved

Discrepancy when data is grouped

  • 11 June 2021
  • 5 replies
  • 117 views

Badge

I generated a below retention report. Starting and returning events are same, but I am segmenting return events based on an additional attribute. The attribute only has two possible values and both of them are used as below.

 

In the report, shouldn’t sum of the return events equal to 100%?

 

icon

Best answer by MikkoKarvonen 11 June 2021, 09:10

View original

5 replies

Userlevel 5
Badge +3

I generated a below retention report. Starting and returning events are same, but I am segmenting return events based on an additional attribute. The attribute only has two possible values and both of them are used as below.

 

In the report, shouldn’t sum of the return events equal to 100%?

 


Isn’t it 100%? 88.19 + 11.81 = 100 and 29.85 + 70.15 = 100?

Badge

I generated a below retention report. Starting and returning events are same, but I am segmenting return events based on an additional attribute. The attribute only has two possible values and both of them are used as below.

 

In the report, shouldn’t sum of the return events equal to 100%?

 


Isn’t it 100%? 88.19 + 11.81 = 100 and 29.85 + 70.15 = 100?

I meant 88.19 + 29.85

Userlevel 5
Badge +3

Retention does not tell you how the users split between the two options you have chosen. It tells you that out of all the users that have done Journey.AddedPhoto at least once, 88.19% have done so again with context = dashboard, and 29.85% with context=my_journey. It does not stop tracking a user once one of those conditions have been met, so if the same user adds photos using both methods, they’ll be counted for both groups. This is why you are seeing more than 100%.

What is the exact question you are trying to answer?

Badge

Thanks for the explanation. I wan to distinguish between the context repeat users engage with. The current report answers the question, but I wanted to double check the Discrepancy as I didn’t expect an overlap

Userlevel 5
Badge +3

Thanks for the explanation. I wan to distinguish between the context repeat users engage with. The current report answers the question, but I wanted to double check the Discrepancy as I didn’t expect an overlap


I would do this kind of chart in event segmentation by setting up a cohort for users who have performend Journey.AddedPhoto 2 or more times, and then chosing the Journey.AddedPhoto event and grouping the results by context. If you set the graph to bar diagram and use Active%, you’ll get a very good picture on how the users split between different options.

Reply