Automated card sort


I need to sort 140 reports into groups based on roles; basically a card sort exercise.

Is there any software that can take data that shows which users access which reports in order to automatically group them?


You mean like a Google Analytics for forms? The first thing that pops into my head is Microsoft Access. But I only have a passing familiarity with the program. As I understand it, Access is Excel on steroids.


I’m not really sure yet. We have several thousand users and 140 reports. They are currently thematically organised but we want to make relevant reports more accessible to users based n their roles.


I basically now have a spreadsheet currently 10,000 rows soon to be 500,000 with 140 different reports in a format like thus:

user ID - Report
001 - report001
001 - report001
001 - report002
002 - report001
003 - report003
003 - report002
003 - report002
004 - report003
004 - report005
004 - report003
005 - report001
005 - report001
005 - report001
005 - report001

I’m already using filters and can view a list of all users who’ve accessed each report by filtering the report column but I was hoping to then create groups of reports based on the fact that the same groups of users are accessing them.

I can imagine something visual with floating clouds of users with each cloud’s visual size determined by the number of reports they have in common. It sounds cool but so niche I can’t imagine anyone has made it.


I think I might be able to help, but I need to know a bit more.
Do you already know the role of the users accessing those reports?
This seems like a simple data aggregation problem, unless you also want to predict the category for future users.
Can you share a bit more?


We don’t know the roles which is the main part of the issue as we were hoping to figure it out based on what they were looking at but I have just persuaded my boss to put a PBI in to ask users next time they log in so I think this problem is solved and it will be a simple case of grouping the reports by the type of user who accessed them.

I guess asking the users wherever possible is always going to be better than trying to guess.


What about looking at groups of users who have many reports in common and separating by that? I think it could at least help to have an initial structure, assigning possible roles related and then validating with the responses. I think many users won’t respond so it’s good to have an initial hypothesis to validate their roles, for example, if you identified 100 users that read the same reports and 20 of them respond then you can group with the rest maybe?
Let me know if this makes sense.