Team, I want to identify both continuous users and ad-hoc users for my product based on a specific event, let's call it X.
Continuous users: These are users who have performed activity X consistently every month starting from their initial month.
Ad-hoc users: These are users who have performed activity X sporadically or in non-consecutive months starting from their initial month.
how can I create 2 cohorts for this purpose ? I tried several things but no luck
Andrew S. will you be able to help me with this one ? its been while I requested this. please . Thanks in advance.
I couldnt find anything similr being answered in other community forums
this is the only thing I could come up with for getting close.
It is finding users who rated a product in 12 distinct months during the past 12 months. (would have to make it 24 if past 24 months). Issue is that it misses users who may have done 6 distinct months in the prior 6 months.
You can do retention graphs like this, which would let you see the users at different points who satisfy the retention, but it's not really what you want.
yes , It cant miss the users who have done continuously 6 months since they login as they have been continuously using the product since then . we are using the retention graphs as u have specified. however I want to see other product metrics against those continuous users vs rest of the product users to compare and contrast. thats why I want to create a cohort for that group , so that i can use it as a breakdown for all other reports on the metrics
can you get comfortable with some sort of shorter time period and use my first example with <12 months. Like who did X at least 1x each month for 3 months or 4 months?
I will try that
thats a good suggestion
The other suggestion, to consider going forward (not backwards), you could have your back end do the analysis for whether a user registered X every month since created_at. And you could then have a boolean for your user profile that passes with your mixpanel data logging for "consistent: true" or whatever you want it to be. if they fall out (did it every month for 5 months but then not the 6th), then they would be "false". You could then do a simple cohort of users where consistent=true.
I created the cohort , however seems its not picking the conditions - did X at least 1x EACH MONTH. my main criteria for continuous users are that they must have performed it EACH MONTH. the final results are similar whatever the criteria used. It just count users who have performed X and not check EACH MONTH.
in the first image change it to 6 not 1
so you want the distinct count of the months = 6 in the past 6 months
your first image is basically finding people who did the event in at least 1 month in the past 6 months