What three years of tickets told me in my first twelve weeks

Twelve weeks into a new job, I read three years of the help desk's own ticket history to understand the place I'd walked into.

I have been at this help desk for about twelve weeks. Not long enough to have seen the history happen. Long enough to read it, though, because every ticket is still there. So I spent some early mornings going through three years of them, back to early 2023.

I did not do it for work. I wanted to know what I had walked into.

Three years is around nine thousand tickets with actual notes in them, plus another eight thousand that have a subject but no writeup. I tagged them by system, bucketed them by half year, and then split them one way that mattered: a person asking for help, versus a machine emailing the queue.

That split is where it got interesting. The single biggest source of tickets over those three years was not a customer. It was one automated document alert. Almost 2,800 tickets in a year, and about 1,300 of them inside five days. Nobody wanted anything. A job failed, a timer fired, a ticket appeared. We still take roughly 85 a month from a job-timeout alert, and nearly all of them need no human at all.

You do not notice that going forward. You just feel busy. It only shows up when you look at all of it at once.

A couple of other things fell out. Migrations clear the board and then leave their own mess. Moving off the old ERP killed about 21 tickets a month almost overnight. A VPN change did the same to another chunk. But each move drops a new set of little problems behind it, and those are in the numbers too. The part that never goes away is integration: trading partners, EDI, certificates, invoices. Twenty-five to forty tickets a month, every period, because every new partner is another thing to wire by hand. And the stuff actually growing is security and identity. Phishing reports are up around sixfold. Onboarding and offboarding climb every time someone joins or leaves.

I did not live any of that. Other people did, and they left it in the tickets. Which is the point I keep coming back to. When you are new, the move is not to show up with a plan and start fixing a desk you do not understand yet. It is to learn what happened before you got there. That is the only way you help steer what comes next without breaking what already works.

I am twelve weeks in. I am not about to start telling people how to run a desk I am still learning. But I know where this one came from better than I would if I had only ever worked the queue in front of me. That is the part I will build on.

← All writing · Portfolio