Do not compare this month’s production spending with an unrelated set of files accepted this month.
As a planning exercise, select a recent cohort and reconstruct its history. If activity records are unreliable, conduct a short prospective time study alongside normal operations. Do not turn employees’ recollection of “about ten minutes” into a precise benchmark.
Which events should the baseline capture?
Create one event log with:
file_id | claim_reference | state | peril | complexity | inspection_source | production_route | event | timestamp | actor | version | active_minutes | defect_code | cost
Record at least:
- Assignment received.
- Evidence received.
- Input completeness checked.
- Missing information requested.
- Production started.
- First complete package submitted for acceptance.
- Review started and finished.
- Return or escalation issued.
- Corrected package resubmitted.
- Production accepted.
- Accepted version successfully stored in the carrier system.
- Subsequent correction or supplement opened.
Keep active work time separate from elapsed time. A file waiting for measurements consumes little estimating time but still remains unfinished.
What should first-pass acceptance mean?
Use a definition approved before measurement:
First-pass acceptance = packages accepted at their first formal acceptance review without a required production revision ÷ packages receiving a first formal acceptance review.
Record minor defects even when the reviewer accepts the package. If the reviewer fixes the work rather than returning it, record that labour and classify the file consistently; otherwise reviewer repair can artificially improve the acceptance rate.
Also report:
| Measure | Recommended interpretation |
|---|---|
| Input-complete rate | Assignments meeting the input specification at the first intake check |
| First-pass acceptance | Performance at the first formal output review |
| Material-error rate | Reviewed files containing at least one defined material defect |
| Assignment-to-usable-file time | Total elapsed time, including evidence waits and corrections |
| Complete-input-to-usable-file time | Production performance after the agreed input gate |
| Review and correction minutes | All relevant touches, not just the first review |
| Exception rate | Files routed outside the ordinary process, with reasons |
| Post-acceptance correction rate | Previously accepted files requiring correction within an agreed observation window |
| Backlog age | Outstanding files by stage, age and owner |
Never improve the reported result by hiding difficult assignments in an unmeasured exception queue. Publish starts, accepted files, open files, withdrawals and exceptions together.
Exit gate: finance and QA can reconcile the cohort, its labour and its status without unexplained missing files.