Retest workflow after rohs reach testing failure: part 4
A credible retest is the last step of a factory correction, not a substitute for one. Buyers should require evidence that the input or process changed, verify that evidence where feasible, scope confirmatory testing around the corrected goods, and release only the lots the new result genuinely covers. This final part of the series shows how to run that closing stretch without letting schedule pressure punch holes in the file.
Key takeaways
- Verification confirms corrected goods against agreed criteria lot by lot before anything ships, which is the acceptance logic behind every retest. [2]
- Sampling discipline requires the retest units to be shown representative through a stated method and reconciled IDs. [1]
- No retest is authorized until the factory produces evidence that the input or process actually changed.
- Scope confirmatory testing around the corrected input with methods equal to or stronger than the original.
- Release only the lots the new evidence covers, holding everything else until its own evidence exists.
The final part of the retest series handles the stretch buyers find hardest: the factory says the problem is fixed, pressure builds to load, and the evidence file still has holes. This guide covers verifying corrective action at the source, scoping the confirmatory retest so its result carries weight, and releasing goods lot by lot against the new evidence rather than by optimism.
Acceptance thinking frames the whole episode correctly. Verification activities confirm that corrected goods meet agreed criteria before they change hands, and each round of testing belongs to a defined lot rather than floating free as a general character reference for the factory. Sampling discipline supplies the companion rule: the units tested must be shown to represent the goods released, through a stated selection method and reconciled records.
What follows moves in factory order rather than laboratory order. First the corrective action gets specified and verified, then the retest gets scoped around the corrected input, and finally the release decision gets drawn with lot boundaries that match the evidence. Buyers who keep this order spend fewer laboratory cycles and defend their decisions more easily.
What corrective action looks like on paper
A usable corrective record names the cause at the material level, states exactly what changed, and identifies the production point from which the change applies. Vague assurances that the factory will pay more attention do not qualify, because nothing in them distinguishes future goods from the failed lot. Ask for the supplier batch records behind the change, the date the new input entered production, and the disposition of remaining old input stock. Acceptance practice treats this package as the precondition for re-verification: the follow-up check exists to confirm a stated correction, not to audition unchanged goods. [2]
Where the stakes justify it, verify the correction on site rather than by email. An independent visit can confirm that suspect stock is segregated, that the new input matches its declarations, and that production records show the changeover at the claimed date. Programs that pair factory follow-up with laboratory coordination, such as UTS restricted substance testing UTS restricted substance testing engagements covering both visits and retests, keep the field evidence and the lab evidence in one consistent story. Remote paperwork alone rarely carries the same weight when a customer later asks what was actually checked.
Set a clear authorization gate: no laboratory run is commissioned until the corrective package reads credibly. This single rule eliminates the most expensive failure mode in retest episodes, which is paying for analysis of goods produced before any correction took effect.
Scoping the confirmatory round
| Scope decision | Recommended approach | What it prevents | Record it as |
|---|---|---|---|
| Suspect component | Dedicated lines, ungrouped from clean parts | Clean neighbors masking the culprit | Line-by-line method note |
| Test method | Equal or stronger than original failure | Downgraded retest presented as clearance | Method citation per line |
| Production window | Only goods made after the correction | Pre-correction stock riding along | Date range on sampling sheet |
| Sampling rule | Rewritten rule reflecting the investigation | Repeat of the plan that missed it | Written rule attached to report |
| Untouched variants | Separate evidence before release | Assumed clearance by association | Explicit hold-or-release per lot |
Running the verification visit well
Confirming segregation and changeover
Walk the hold area first and confirm the quarantined lot is physically intact, labeled, and excluded from packing lines. Then trace the new input from receiving records through stores to the line, checking batch numbers at each point. Photograph labels and record numbers rather than accepting summaries, because the retest later claims to represent exactly these goods.
Witnessing the fresh sample draw
Draw the confirmatory samples from post-correction production under the rewritten rule, spreading selection across cartons and positions as sampling discipline requires. Reconcile unit IDs with the laboratory upfront so the intake record and the field sheet describe the same units. [1] This reconciliation is the detail reviewers check first when a retest is challenged, and it costs nothing to get right at the time.
Releasing goods against the new evidence
- Map lots against retest coverageList every held lot with its production dates and input batches, then mark each one covered or not covered by the retest sampling.
- Release strictly inside the evidenceRelease only lots whose input, dates, and sampling all fall inside the retest evidence, keeping the boundaries explicit.
- Hold everything outside the evidenceKeep pre-correction and untested-input lots on hold pending their own sampling and results.
- Memo the decision with linksWrite the release memo linking the corrective record, both sampling sheets, both lab reports, and the lot map.
Corrected goods earn release through verification, not through elapsed time. Hold the authorization gate until the corrective evidence reads credibly, then let the confirmatory result speak only for the lots it genuinely covers. [2]
Sources
- Wikipedia: Acceptance sampling —— Overview of acceptance sampling including AQL concepts used in product inspection.
- Wikipedia: Factory acceptance test —— Overview of factory acceptance testing performed before shipment or installation.
Frequently asked questions
What must the factory show before I authorize a retest?
Evidence that the input changed, not promises that it will. Expect updated supplier declarations or batch records, photos or records of segregated nonconforming stock, and a description of the process control that prevents recurrence. Schedule the retest only after this package arrives and reads credibly, because testing goods produced before the correction proves nothing.
How should the confirmatory retest be scoped?
Test the corrected input first in isolation if possible, then test finished goods from the new production under the rewritten sampling rule. Keep the suspect component on its own report lines rather than grouped with clean parts. Match or exceed the original methods so the new result can genuinely supersede the old one.
Can I release part of the held goods?
Release only the lots the retest evidence covers: same corrected input, same production window, same sampling rule. Lots produced before the correction, or from untested input batches, stay held until their own evidence exists. Partial release with clear lot boundaries beats both blanket rejection and blanket acceptance.
Why not just retest immediately and skip the factory follow-up?
Acceptance practice distinguishes verification of corrected goods from repetition of the same check on unchanged goods. The first resolves the finding; the second merely spends budget while suspect stock sits. Require the corrective evidence, verify it where feasible, and only then commission the laboratory run.