our 224 brackets have more passes than accepted items

Yes. The labels persist and quality decisions are saved. The dashboard rebuild adds successful attempts, so a bracket with two successful checks contributes twice.

Count the current accepted brackets, and keep attempts as a separate measure. I would also show what happened to the others. A correct accepted total can still leave the planner guessing whether the remainder failed or has not been checked.

Try the rebuild on the saved batch first, because you already have the identities and decisions to compare with, rather than buying a new dashboard to count the same wrong thing more attractively.

The saved-batch comparison gives 209 currently accepted, nine awaiting a reviewed retry and six not yet inspected. That accounts for all 224. Attempt history is unchanged.

Does a withdrawn acceptance reduce the item count once, including after reopening?

We tested withdrawal and restoration with quality, plus repeated result delivery and reopening. Each changes the accepted-item count only when the decision changes. The installed report now matches the checked batch list. Thanks, Aisha, the remainder categories caught a misleading 'nearly finished' heading too.

That is a report I could explain to a covering tutor. The nine awaiting another decision no longer disappear into the same bucket as six untouched brackets.