The FR5 acceptance dashboard looks nearly finished after restart, but repeated inspections are inflating its count against our 108 labelled housings. Failed checks can have reviewed retries. How do I count accepted items accurately while retaining every attempt and its outcome?
Build the accepted total from identified items with an authorised accepted disposition, not the number of completion messages. Keep attempt records linked beneath those items.
And agree what establishes that disposition. A failed attempt followed by any later completed check isn't automatically an accepted housing; the retry result and review still need their proper meaning.
Item labels are available and retries have separate review records. The dashboard currently increments on each completed inspection, which is why repeated attempts look like extra production.
What happens to an attempt with a missing or unreadable item label? It shouldn't be assigned to the nearest housing just to get the reconciliation finished. Leave those rows visible for someone who can actually resolve them.
Yes, unmatched stays unmatched. Also test replaying one completion twice and restarting with the real saved attempt history, not an empty convenient counter.
I've drafted an item-level count with unresolved identities separate and failed attempts retained. Quality is confirming the disposition rule before we recalculate the batch. No manual reduction of the counter to a guessed number.
Preserve the old reported total with the reason it was wrong. Correcting the dashboard should make the audit trail clearer, not quietly replace the original report and leave people unable to explain why progress went backwards.
Will the ordinary screen show both accepted items and the unresolved work? I'd rather see a smaller honest total than one that makes the next shift think only a couple of housings remain.
That is in the screen draft. Recalculation and restart tests are pending the confirmed disposition rule; the old total is marked attempt-based and unsuitable as batch progress, with its history retained.
Include an item failing then passing a reviewed retry in those tests. It should contribute one accepted item while still exposing both attempts and the review.