When Everyone Skips the Process, Fix the Process

There's always a worn dirt path cutting across the grass where the sidewalk makes you go the long way around. Nobody plants it. It just shows up, because enough people made the same decision on their own.

We had one of those at work with an approval chain — four steps on paper, two in practice. I found it doing something as boring as reviewing old tickets, and the first version of the workflow diagram matched nothing anyone was actually doing.

My first reaction wasn't great. Some part of me wanted to just point at the doc and say do it this way, it's not that hard. I've built enough of these over the years — incubator, state programs, plenty of companies since — to know that reaction is almost always wrong, but it still shows up in me before the better one does. I caught it that time. I don't always.

What I actually believe, and have believed since long before I had a name for it, is that if nearly everyone routes around a step, the step is the defect, not the people. I'd been running some version of this logic since I was fixing jet engines — there, the tolerance was the tolerance, no working around it, because the cost of skipping it was obvious and immediate. Office process doesn't announce its cost the same way, so people just quietly stop doing the part that isn't earning its keep. That's not defiance. That's data.

The part I'm still not great at is telling that apart from the other kind — where the shortcut is the actual problem and the process was right. I don't have a clean test for it. Sometimes I ask why the step exists and nobody remembers, which tells me plenty. Sometimes I ask and get a very good reason I'd forgotten, and I feel a little foolish for assuming the room was smarter than the document by default.

I lean toward trusting the workaround first and checking my ego second. I think that's mostly right. I'm not fully sure it always is.

If almost everyone skips a step, how do you decide whether the step is broken or the skipping is?