Total Number of Users Over Time and Percentage of Users in Drill

I'm using countly enterprise v18.04.1.

I'm wondering if can do some of the things I'm asking. If it can't, then are there any plans for to do these things?

Is there a way to get the number of users at a point in time or in a bucket of time with provided filters?

So Is there a way to get a graph that shows the total number of users each month over the period of a year?
Then could we filter that by any property on the user to get that total?

Also, could that data be combined with the event data in drill to get the percentage of users that completed a particular event in a given month?

I had the idea of having a users activated event that would have a sum of 1 when a user was activated and if they were deactivated the event would have the sum of -1. Then you could run a drill query to get the net users activated in a month. The only issue is I want a rolling balance of total active users month after month ( or day/hour/week). Is there a way to do running totals in drill with sums?



  • Hello,
    so here would be drilling event by sum, with monthly buckets:

    Does it look like what you need?

    Comment actions Permalink
  • Hey Artus,

    I'm wondering if countly can do running totals on sums. Let's say I started collecting data for countly in January - I got 10 users. Then in February I got 14 users, March 20 users. But let's also say users can be deactivated. So in February 2 users were deactivated, and March 5. Let's say I make an activation event to track activations and deactivations, so sum of 1 for when a user is activated and sum of -1 for when a user is deactivated. Right now in drill you can get the data like this:
    Jan: 10
    Feb: 12 (14 - 2)
    Mar: 15 (20 - 5)

    But what if I wanted that shown as a rolling total? Let's say I wanted to see the total active users I have in any given month, like this:
    Jan: 10
    Feb: 22 ( 10 + 12)
    Mar: 37 (10 + 12 + 15)

    Are there any plans to support something like that? The tricky part is I want to jump in at an arbitrary time - say like Feb & Mar and still have the totals right:
    Feb: 22
    Mar: 37

    Also, It would be awesome if I could filter by customer user properties and event segmentations as well.

    Does do something like this currently?

    Comment actions Permalink
  • Hello,
    I don't think there is anything like that available out of the box.
    Specifically I don't think we could do such cumulative calculation by month, increasing number for each next month (starting I assume from January).
    Mostly because if you select period Feb and March we only read in data about Feb and March and work with that in terms of calculation and segmentaiton.

    That could only be achieved if you resync active users on the 1st of each Month, so on first of Feb, you resync 10 users from January, so their events are stored in February too. Then something like this could work.

    Would it be possible for you?

    Comment actions Permalink
  • Just in case someone is looking at this thread in 2020 :) This kind of reporting is available in the new Cohorts UI in a way. 

    If you create a cohort that groups users based on;

    - Performed activationEvent "All time"

    - Didn't perform deactivationEvent "All time"

    It'll give you the change over time in number of total users like below, for users who activated but never de-activated (notice that the cohort is based on all time range, but can be adjusted to any day scale as you need).

    You can view this directly in Cohorts or you can add Dashboards widgets to have custom dashboards built around this metric.

    Comment actions Permalink

Please sign in to leave a comment.