Feedback turnaround and unanswered work
Measure returned feedback separately from open submissions so unfinished work does not disappear from the service record.
Begin with a complete submitted population
A turnaround summary needs a declared list of submissions and a clear starting event. This exercise uses the supplied submission timestamp, not payment date, first chat message or an inferred review start. Keep every task identity, including pending and withdrawn work. A provider may use different contractual definitions, which require separate source evidence. The sample packet is authored and describes no real mentor. Its five tasks show why a neat average can hide unfinished work. The journal cannot detect omitted tasks, authenticate timestamps or establish that all submissions were accepted. Those source questions remain part of an actual service review.
Returned and pending time are different
Returned turnaround runs from submitted to returned. Pending age runs from submitted to the as-of timestamp. Pending work is not assigned a fictional completion date and does not enter the returned-only mean or median. In the base teaching packet, three returned tasks take two hours, twenty-four hours and one hour. Their median is two hours and mean nine hours. Another task remains pending for fifty-one hours at the declared cutoff. Excluding it from completed duration does not mean excluding it from the record. Its separate age and status remain visible so a summary cannot pretend the slow unresolved item never existed.
Declare the deadline operator
A due time and its endpoint convention must be entered explicitly. Under an inclusive deadline, a response exactly at the due time is on time. Under a strict deadline, equality is not on time. The same convention defines when an open task becomes overdue at the cutoff. The strict variant changes an equality classification without changing the actual timestamp. This is a teaching interpretation, not a provider service-level agreement. Keep the agreed term separately from the supplied due field. A result under an entered operator cannot settle a dispute about what the agreement promised, which clock it used or when a submission was accepted.
Missing due times stay unclassified
A returned task with no due time still contributes to the duration population, but it cannot contribute to a deadline comparison. The journal lists its identity separately. A withdrawn task has no completion or open-age claim in this exercise. Its presence remains visible without inferring who withdrew it or whether a fee was refunded. Those questions need additional records. The sample's F4 returns in one hour without a supplied due time; F5 is withdrawn. Neither should be silently relabelled a timely completed review. This distinction prevents a service summary from manufacturing deadline coverage that its inputs do not provide.
Review the distribution before the slogan
Mean and median describe the returned population only. A fast median can coexist with a late returned task and old pending work. Read the returned count, open count, unclassified deadline records and observed cutoff together. When the pending sample later returns, the returned population and its statistics change. That is a new scenario rather than a correction of the earlier snapshot. The tool does not infer customer satisfaction, critique quality or strategy improvement from speed. A response can be prompt and irrelevant, or delayed and technically useful. The bounded result is a reproducible service-timing record with explicit missing coverage, not an endorsement or a typical-provider claim.
The cutoff changes the open age
For fictional F3, a cutoff at 10 October 12:00 UTC produces fifty-one hours of pending age from its 8 October 09:00 submission. A later cutoff would change open age without creating returned feedback. A later actual return would instead move it into the returned population. Those are different observations. Preserve the old snapshot and its as-of timestamp. The journal cannot choose the convenient cutoff that makes a service look faster or invent a return to improve the average.