For two years my team's performance was reported upward as tickets closed per week and average time to close. Both numbers improved steadily throughout that period. So did the number of incidents, the backlog of known faults, and the volume of the same request arriving every Monday from the same department. Nobody upstairs noticed the contradiction, because the report contained the first pair of numbers and not the second.
The behaviour it produced was completely rational. A recurring password sync failure took about six minutes to resolve by hand and appeared roughly twenty times a week. Actually fixing the underlying directory replication problem was a two-week piece of work with a real chance of making things worse before it made them better, and it would have removed twenty closures a week from our figures forever. There was no line on any report where that trade looked good for us. So for two years we resolved it by hand, twenty times a week, and were commended for our throughput.
I do not think anyone designed this maliciously. The metric was chosen because it was easy to extract from the ticket system and because it sounded like productivity. That is how most bad metrics arrive: not from a bad theory of the work, but from the absence of any theory at all, filled in by whatever the tool happens to export.
What changed things was not arguing about the metric in the abstract. It was putting a second number next to it. We started reporting repeat volume, meaning how many of this week's tickets were the same underlying issue as something we had already closed, and we split the backlog into new causes and known causes. The moment those two lines sat beside the closure count, the conversation upstairs changed on its own. Nobody had to be persuaded that eliminating a recurring fault was valuable. They had simply never been shown that it was possible.
If you want to know what a team will actually do, do not read the strategy. Read the report that goes to their director. People optimise what is visible, and if the only visible thing is motion, you will get a great deal of motion.
– Serguey Shinder
Top comments (0)