Every improvement programme gets a name. A logo. A launch. The ideas inside it were already there, they just needed a consultant to present them before anyone would listen. The programme delivers. The posters fade. The results slip. The consultants return. Nobody asks why.
An operator raised a problem at 5am. Within one day it was solved. Three weeks later I watched the same instinct fail completely, because the system to catch it did not exist. Same people. Same problem. Different system.
I've asked this question in a lot of plants, to a lot of people. The answers are almost always the same. Not wrong. Not evasive. Just untethered from anything you could verify in the moment.
OEE cannot exceed 100%. So when I found 106.4% on a packaging line, I went looking for the error. What I found instead was a system designed to limit questions rather than surface performance.