Sixty-two audits in, the method is stable but the failure modes are not gone. These are the four that recur, and what we changed to catch them earlier.
We watch the loud queue
The team points at the workflow that generates complaints, and it is rarely the one that costs the most hours. We now time every workflow before ranking any of them, and only then look at what people flagged.
We take the process document seriously
It describes what should happen. On a first audit we once mapped six steps that nobody had performed in two years. The check is simple: nothing enters the map unless we have watched it happen.
We under-ask about exceptions
Ninety percent of a workflow is easy to see in ten days. The ten percent that decides where the handoff rule sits lives in a person’s head, and it comes out only when you ask what the weirdest case last month was.
We schedule the handover too tightly
Ninety minutes is enough to walk the plan and not enough to argue about it. We now book two sessions, and the second one is where the build order actually gets decided.
002
/ Keep reading
From the journal