We identify "active users" as a user who has done a thing in the last 90 days. But using this kind of cohort means if we look back longer than 90 days, the criteria is not relative to the time period but still in the most recent 90 days... For example "how many active users did we have 12 months ago?" we cannot get this because it will look at users 12 months ago, who have completed a mission in the current "last" 90 days. What am I missing to be able to make such a cohort and have it be usable for all time?
Oh really?
Give it a try, let me know what you get.
well I have been doing this, but only todays thought I realized the problem... Maybe you can help me because I am very confused. These two graphs "should" show similar stuff
The graph on the left should be both "activated" and "non-activated" bees together. But it represents a totally different curve
So the results you’re looking for is the one on the right, the one on the left shows different results because: 1. it seems like the cohorts are different 2. I wish i knew why it’s like that but it’s just the way cohorts work, if you want to see how many users are in the cohort at any given time you plot it overtime.. when adding filter it seems as though just the latest period makes sense
oh...
Yes that makes sense, I am adding it as a filter
Right the graph on the left is using a cohort as a filter and the graph on the right is graphing the cohort over time
this 3-month hump makes perfect sense and thats what got me on this panicked line that cohorts don't work
