The data usage varies widely with tens of factors in an implementation, including but not limited to device platform, enrollled mode, firmware version, connection mode & schedules, policies such as profile assign/revoke schedule,  app deployment/uninstallation, file-sync/data-collection rule schedule, use of TEM and or firewall/restrictions policy, ..., etc.   If an implementatiion has a certain maximum target budget, it is the MDM administrator who should select relevant configurations to achieve the target, possibly with different levels of sacrifices on responsiveness, visibility, controllability, security, etc.

If you can get document on  MAXIMUM data usage or commitment from any MDM software vendor on the market, please share the information.  

I'm interested in this topic as well.  We've recently deployed MC into an IOT Linux solution with very low data requirements (sends data 3-5 times per day) over cell network and data consumption has went up drastically since deploying with MC.  These are very low used devices as far as MC is concerned:  no profiles or app deployments, file sync rules, etc....  We are really only deploying into MC so we have remote ability to update firmware as needed in the future.

I am looking at paring back device check-ins for the Linux devices from 2 hours to once a day (or week) but I really don't have an idea of what impact that will make.  Is there any detail on how much data is used for a device check-in per device type or similar as I would think that would be fairly standard?  I have also thought about changing the connection from persistent to 9-5 or something like that.

It doesn't sound like a lot, but 20-30MB of usage is a lot of data in a low data IOT solution where we have very competitive cellular rates with carriers based on a very large number of devices and low usage.