Am I automating the easy five seconds and keeping all the faff?

More than half our attended time on the mixed box. I left it outside the first comparison. Very flattering comparison, that one.

Then rerun the comparison from the same incoming condition on both methods; the robot version cannot be handed prepared bags while the manual version gets the box as delivered

Agreed. Manual operator takes them straight from the incoming box. My robot tray starts looking like a product photograph. That's the missing work.

And which bag type makes most of the preparation work? The mixed total might hide one order that behaves reasonably well.

I would be cautious about cancelling all of it from the mixed trial, if one stable order represents enough work to justify a simpler arrangement. I cannot tell whether you have such an order, though.

A narrower order still needs its own costs, including changing into and out of that arrangement; otherwise the small successful corner becomes another flattering comparison

One repeat bag is much easier. Still needs orienting. Maya, another operator can run a prepared tray but calls me when the corners won't lie down.

Then don't count your absence as free operator time. They still need your help, even if you answer from another bench.

Could that operator prepare the repeat order and record every time they need you? Not to catch them out. To find the work your hands are quietly supplying.

I would make that a bounded comparison, with an agreed decision afterwards. Continuing to develop inserts while measuring their benefit makes it difficult to know which arrangement you have actually evaluated.

I'm tempted to make one shaped tray first. Current one is flat and clearly wrong for the bag. Feels daft to cancel without trying the obvious fix.

Price that trial before defending it; 'one shaped tray' can include design, manufacture, storage and another changeover routine, not just a piece of plastic

What would make you stop after that tray?

If another operator still needs me to rescue ordinary bags, or whole-order labour isn't better than manual. We need to agree how much improvement is worth the maintenance burden.

That last point matters. A tiny labour saving may be unattractive with frequent changes, but there could be another benefit worth discussing, such as reducing a tiring repetitive task. Have the operator weigh in on that.

Do they actually prefer using it?

They like not reaching across the tray. Hate fiddling with seams. Useful distinction, thanks. I've asked them to compare both jobs, not rate the robot as a whole.

Have you agreed the trial budget, or is the shaped tray already being made?