We once had a report that counted the later pass correctly but hid the earlier failed attempt. Keep the attempts as history and derive the item disposition separately, with quality owning which reviewed result governs the current decision.

Also check whether an accepted housing was later placed on hold. The most recent success is not necessarily the current disposition, especially when someone reviews an earlier result after the inspection itself.

The housing labels survived restart. I have 44 attempt records for 40 housings, plus eight with no attempt recorded. The dashboard had been treating 44 as progress towards 48 accepted.

So even distinct inspected housings would be the wrong accepted total; quality still needs to reconcile those forty dispositions.

Quality's first reconciliation is 35 accepted, three on hold and two rejected among the forty. Four repeated attempts remain in history; none creates another housing.

Keep the eight with no recorded attempt visible as their own group. That label says what the records establish; it does not by itself prove nobody inspected them or tell the shift to run them again.

Does rebuilding the report from the saved records reproduce those same groups?

Test a changed disposition too. A housing moving from accepted to hold should leave the accepted count even though its successful attempt stays visible, and reviewing a retry should not alter the batch membership.

The revised report rebuilds 35 accepted from the saved records. A test moving one accepted housing to hold produces 34, with its pass still shown. Reopening the report keeps both the decision and its review trail.

Has quality confirmed the eight-record gap, or is that still a separate reconciliation job?

The eight are confirmed awaiting inspection. Quality signed the current breakdown: 35 accepted, three held, two rejected, eight awaiting. All 48 accounted for; the batch itself is not finished.

And a repeated delivery of an existing attempt leaves that count unchanged?

Yes. Replaying an existing attempt and reopening twice changed neither totals nor dispositions. The reporting fault is fixed, and the remaining work lists are now separate from the attempt history.

Keep that distinction in the shift summary too. Thirty-five accepted is the useful output figure; forty-four attempts explains work performed, including retries, without pretending those retries produced additional parts.