Issue: recover partial results of a bulk ticket assignee change.
Scope: assignee only; no deletion, refund or customer messaging.
Input: selected 100 / completed 90 / failed 7 / unknown 3. Failure means confirmed non-application.
Decision: no retry-all. Exclude successes. Look up original results for unknowns.
Retry: resolve the cause and recheck current permission, business conditions and version; confirm subset and intended change. Link a new request to the original.
UI: retain total counts; state filters; exclusion reasons; latest/intended assignee; last-check time; task-history re-entry.
Records: request scope, actor, item results, before/after values and linked attempts. Do not collect message bodies by default.
Decide before launch: maximum selection, refresh interval, duplicate scope/lifetime, result retention/access, unresolved-outcome deadline/owner/alerts.
Block launch: repeated successful items, unknowns relabelled as failures, conflict overwrite without reconfirmation.
Before adoption: integration QA in the real test environment and responsible-role confirmation are required.