Dynamically Identifying frequency of errors in logs

I will start out by staying, I am not a developer. My coding skills are elementary so I apologise if this has been answered in another place and I missed it.

Our software, which runs on customers machines directly, generates and uploads "error" logs. Presently, these logs just get uploaded to a folder on our server, and once every 30 minutes or so, a service reads them for the customers ID and moves them to the customer's account. There they sit until someone manually looks at them and decides if they mean something and what to do, if anything.

This works well for retroactively looking to see if the customer has the same issue as another customer.

What I would like to do is have the service that reads the log for the customers ID also use some sort of pattern analysis to identify each unique error and then record the frequency over time. Such as 1 day, 3 days, 5 days, 1 week, 2 weeks a month, etc.

Part of the issue is that the error logs dynamically generated on the customers' machines. Some of the errors are prewritten by us in code, but many of the errors we get come from 3rd party sources and integration. So many of the errors we get are not something we would know to look for ahead of time.

A good example of this, Dell changed one of their system tools that started deleting several of our files; it generated a specific error. Not one we wrote, but still specific enough that once we identified it, we could track it.

So, after that long winded explanation. Is there a good way to do this type of analysis? We typically prefer using the .net frame work as much as possible. Most of the results I see depends on looking for an existing pattern, which isn't quite what I am looking for.

Thank you;