I handed operators a saved FR5 tray-to-fixture job and they can start it, but an empty housing pocket or interrupted run brings them straight back to me; that sounds like decisions I never handed over, not people refusing to learn. How do I find the missing part without writing a book nobody can use?
Watch one of those handovers with the operator describing what they can establish at each decision. You may be relying on a physical observation or history that isn't visible to them. The missing part is not necessarily another paragraph.
That already rings true: I recognise which tray was in place before the interruption, while the screen just shows a pocket number. An operator coming back from break can't establish that from what I've given them.
Then the recovery design needs to account for lost or changed tray context, with an authorised response when it cannot be established. I wouldn't teach people to infer the tray from a remembered pocket number. Review it with whoever owns the cell's recovery behaviour.
The cell lead and I have added that uncertain-tray case to the recovery review, and operators walked through the draft without me supplying the missing history; it exposed another assumption about who records a tray exchange, so that still needs a design answer.
Useful that the walkthrough exposed it before you called the handover finished. Keep the unresolved exchange-recording question attached to its owner, and repeat the operator walkthrough once the response is agreed; readable instructions can still describe an impossible decision.