Changeover was treated as tribal knowledge. The line had the signals — the last good piece, the tool that drifted, the order that always stalled — and none of it was in the plan.
We put those signals next to the schedule the supervisors already run. The model does not run the line. It tells the crew what to stage before the current job ends.
Changeover on the primary line ran 18% faster through the quarter, without a new system for people to learn on a Monday.
What we did
- Sat with the crew through a full changeover and wrote down what they actually check.
- Joined machine signals to the schedule, and ignored the rest.
- Delivered the next setup as a short list inside the tool the floor already opens.