Frequency on the Account Level
Hello!
I'm hoping to gather data on the avg number of logins per month per account.
I'm struggling to create a metric that I'm confident in.
Also, would like to show a breakdown by visit numbers (the freeform attached side note - a quick segment showing logged in state.
For my metric, I'm getting an avg of 2 when i breakdown by month, which is leading me to think I'm wrong.
Would love some ideas/ feedback on getting this sorted so I can feel confident in the accuracy of what I'm reporting on.
Ok, let's take a look at this.
The Round function consists of the metric you want to round, and the number of decimal places you want to round to... Using Approximate Distinct Month as the metric, and rounding to your instances of account numbers probably isn't doing anything that you expect.
The Approximate Count Distinct Month might be useful to get a denominator for a variable month report... but let's put a pin in that for the moment.
The next thing I see if the use of "Visit Number" which is the user's all time visit number from the beginning of time (so long as Adobe recognizes them), this doesn't really make sense in the context of what you seem to be trying to pull.
So let's try and tackle what you are trying to build:
"avg number of logins per month per account"
Though, ow your site works does warrant asking: does your system allow users to stay logged in for a long time.... meaning if users log in, and don't log out... they will be counted as users each month, but not as an "actively logging in" action..... Or are you trying to find "logged in" vs "login actions"? There is a big difference.
So I have a similar setup in my account.. I started by using your Approximate Distinct Count for Months (to try and ensure the denominator reflected whatever time frame my report is looking at... If I break this down by Month, then I am essentially dividing the Login Success by 1 (at the month level), but on the roll up total, that is where I will see the magic of the monthly average.
Now, in my case, users can stay logged in, so if I go back, and nest the previous division in a container, then add a second division using the distinct count of my user ids:
This should get me Successful Logins, per month, per user.
My average per month is less than 1 (which is expected for us)
However, if I really wanted to see "avg logged in users" (i.e. not dependent on the login action).
I track a login state on all pages, and I I have a segment that looks for the login state to be true. In this case, I would probably create a simple calculated metric of "distinct user ids" / "approximate count month" (again to ensure that if my report is 12 months or 3 months, the total of the column takes the range into account), then I would just drop my logged in segment on top.
You mean the average "logged in users" as opposed to "average successful logins"?
Mind you, this will only work if you are tracking a login status.
So hypothetically, if you were doing that (on every page view and action) tracking a yes/no or true/false status for logged in). I created a hit level segment looking for logged in to be yes or true.
Now, to get average User ID per Month you should be able to do something like this:
Keep in mind, that users, IF you remember their User ID when their session time out (like I do) will show in both logged in and logged out segments (i.e. at different points in the month and the year, the users will have different statuses, so if you add logged in and logged out together they will be higher than the total)
I would use this as a basic calculated metric, then stack my "logged in users" segment on top of it.
Now, if you have a lot of the same users each month, you will notice that when doing a breakdown by month, your average at the top is going to be a lot less than each month... let's assume that over the course of a year you have 500,000 user ids, but most of those are repeat users... if you divider 500,000 by 12 months, your average is about 41,000-42,000
Now, maybe this isn't exactly what you want... maybe you have numbers over 3 months like:
400,000
500,000
300,000
and you want the total to come out to the average of those numbers: 400,000.
IF you are doing a monthly breakdown, instead of dividing by the number of months, you can use the Mean function.... but this will create an average based on the rows... so it will ONLY work if you are doing a monthly breakdown
But I wasn't sure if this would work for you, based on your freeform screenshot... but this would get you a better overall monthly average.... that would not de-duplicate your Users....
I am not sure offhand how to replicate this average user number without the monthly breakdown..... However, you could in theory (assuming you don't need to use your breakdowns for other things) just collapse the months.. this would mean you could allow Adobe to calculate the average for you (which is probably a better, and more expected number)
Related Conversations
