I Could Be Wrong

As an IT professional, who supports a highly integrated CRM system, I find that it is my system that is the front line with the users. They understand (well, most of them do, anyway), that there are other systems involved, but if they start with my system and an integrated piece throws an error, they see it as my error. I see it as my problem. Like you, my approach is sometimes to direct them to the appropriate person, sometimes educate them , and sometimes I just manage it myself – forward the problem to the relevant person, who may be from a different vendor organisation, and follow up to resolution.
After all, the user doesn’t care who fixes the problem, only that they get it fixed.

One thing I have found is that I almost always blame my own code first, and go looking to see what I might have done to cause the problem, even if I haven’t changed anything. That approach has led me to uncover some as yet undiscovered bugs in my own code, and fix them before anyone notices. I guess looking for problems that aren’t there can cause you to check for scenarios that you thought may never happen, or to look at your code in a totally different way.

I recently had an error reported to me, that I recognised as originating in one of the integration pieces we have. I knew I hadn’t implemented anything recently, and the same piece of code had been working the day before. We had recently made some environmental changes, and of course we all immediately suspected that this was the cause of the problem.
The administrators did their own investigation and found no problems. We had all hit the wall. Since the error message was coming from their piece I could have left it at that, passed on to them for resolution, but with integrated components sometimes it is not that straightforward…. and the user was still unable to perform the task. The error message didn’t make any sense in the context of what the user was doing at the time so I tried to replicate the problem in a different environment. Fortunately I could, and with some logging on I discovered that the error was in my code after all. The error message was totally misleading and was caused by a previously unreported error, which was caused by a particular (and uncommon) dataset. It took almost 2 days to find, but if I hadn’t persisted with investigating the mystery it could have taken a whole lot longer.