Skip to content

Editorial · Fashion Model BCN

Red flags commonly noted in quality management system audit: part 2

Fashion Model BCN
The short answer

Beyond the familiar paperwork findings, auditors raise a second family of red flags around sampling logic, acceptance discipline, traceability, and subcontracted work. These flags matter because they decide whether the factory's good-looking records describe products that actually conform. This second installment explains each flag the way an auditor sees it, with the practical checks that remove it before the visit.

Key takeaways

  1. Sampling plans that cannot detect process defects make every later quality record unreliable. [1]
  2. Acceptance reasoning borrowed from equipment commissioning shows whether the factory proves conformity before release. [2]
  3. Traceability breaks between incoming lots and finished goods are found by simple backward matching.
  4. Outsourced processes sit inside the system boundary and need their own controls and records.
  5. Process monitoring records must show real values with reactions to deviations, not copied ticks.

The first part of this series covered the classic management-system flags: uncontrolled documents, hollow corrective actions, expired calibration, and disengaged people. This second part moves downstream into the technical evidence auditors probe once the paperwork looks orderly: how the factory samples, tests, traces, and controls the processes it does not perform in-house. These are the flags that surprise production managers, because the procedures often exist while the underlying logic has quietly decayed.

Sampling and acceptance decisions deserve special attention because they determine which defects ever reach the auditor's attention. A factory can run tidy meetings and keep neat logs while still shipping mixed lots, because its sampling plan cannot detect the problems its process creates. Auditors trained in product verification look for exactly this disconnect between administrative order and technical control. The findings feel technical, but their root cause is usually the same management inattention as the simpler flags.

The four areas below follow the sequence of an auditor tracing product backwards: from the sample that should have caught problems, through the acceptance reasoning, into the traceability chain, and out to the subcontractors beyond the factory gate. Each section describes the observable symptom, the test the auditor applies, and the practical fix a factory can implement before the visit.

Sampling plans that prove nothing

The sampling red flag appears when the auditor compares the factory's sampling plan against the defects its process actually produces. A typical case: final inspection pulls a handful of pieces per carton against a printed defect list, while the dominant failure mode is an intermittent electrical fault that only a powered functional test would catch. The sample passes month after month, customer returns accumulate, and nobody connects the two because the plan was copied from a previous product years ago. Auditors spot this by asking which defect types each sampling step is designed to detect and finding no answer.

Illustration: Sampling plans that prove Decorative illustration for the section "Sampling plans that prove"; visual only, carries no data.

Sound sampling rests on statistical reasoning that many factories never revisit after the first plan is written. Published references on acceptance sampling explain how sample sizes, acceptance numbers, and lot definitions interact, and why a plan that ignores lot homogeneity or defect clustering gives false confidence. [1] The auditor does not demand textbook statistics from operators, but does expect the quality engineer to justify sample sizes, to define what constitutes a lot, and to show that plans tighten after failures rather than staying frozen.

The practical fix starts with one product family: map its real defect history over six months, check which inspection step could have caught each defect type, and revise the plan so every recurring defect has a named detection point with a defined sample. Keep the before-and-after analysis as a record, since it demonstrates the corrective logic auditors want to see. Factories that do this exercise often discover their inspectors were diligent and their plan was blind, which is a far cheaper problem than the workforce issue they feared.

Acceptance logic that skips the proof

The second technical flag concerns acceptance: goods move from incoming to production, and from production to dispatch, without any recorded verdict that criteria were met. Pallets arrive and go straight to the line because the supplier is familiar; finished lots ship because the schedule is tight and last month's lots were fine. The forms may even get signed afterwards, which converts acceptance from a decision into a ritual. Auditors test this by selecting a recent lot and asking to see the acceptance evidence that existed before it moved, not the signature added after the question.

Industrial practice outside consumer goods shows what genuine acceptance looks like. Heavy equipment buyers run structured acceptance tests at the manufacturer's works before shipment, with agreed procedures, witnessed results, and formal sign-off that conditions release. Overviews of factory acceptance testing describe exactly this discipline of proving conformity where correction is still possible. [2] A factory that applies even a simplified version of that thinking to its own incoming and final gates, with criteria, records, and hold authority, stands apart from one that waves goods through.

Good looks like visible hold points: incoming lots staged with status labels until the verdict, finished lots blocked from loading until final results are recorded, and named people with the authority to stop movement. The auditor will ask who can hold a shipment and when they last did so; a credible example from recent months answers the whole line of questioning. Where schedule pressure previously overrode quality holds, log that history honestly and show the changed rule with its first enforced case. Teams preparing for closer scrutiny often arrange for UTS to observe a trial acceptance round, which surfaces procedural gaps while they are still cheap to fix.

Trace directionWhat the auditor pullsTypical break foundConvincing evidence
Backward from finished lotCarton numbers to component lotsComponent lot never recordedLot linkage on traveller or barcode scan
Forward from incoming lotReceiving lot into finished lotsSplit lots with no split recordSplit and merge log with quantities
Sideways into reworkReject bin into repaired unitsReworked units lose identityRework routing cards with retest results
Outward to subcontractorPO to subcontractor batch recordNo batch record returnedSubcontractor report matched to PO and lot
Across calibrationTest result to device certificateDevice used past due dateCertificate current on the test date
Illustration: Trace direction What the auditor Decorative illustration for the section "Trace direction What the auditor"; visual only, carries no data.

Subcontracted work outside the system

Outsourced processes form the final flag in this group and often the most uncomfortable one. Plating, heat treatment, embroidery, firmware loading, or final packing happens at a subcontractor's site under nothing more than a price agreement, with no specification of requirements, no incoming verification of returned work, and no evaluation record for the provider. The factory's system boundary quietly stops at its own gate while its products do not. Auditors treat this as a scope hole rather than a purchasing detail, because the customer receives the subcontracted quality as part of the product.

Closing the hole does not require auditing every subcontractor on site before the visit. It requires the basics for each outsourced process: defined requirements communicated to the provider, evidence the provider was evaluated and is re-evaluated, identification of subcontracted lots, and verification of returned work proportionate to risk. A simple file per provider holding the evaluation, the requirements sent, and recent verification records resolves most findings. Where risk is high, schedule the provider visit and log it as a planned action with a date.

The pattern across all four flags is identical: the system stops one step before the evidence would become uncomfortable. Sampling stops before the defect it cannot catch, acceptance stops before the hold that would delay shipment, traceability stops before the messy split lot, and control stops at the subcontractor's gate. Auditors are trained to walk exactly one step further than the records invite them to go. Factories that walk that extra step first, and fix what they find there, convert the audit from an exposure into a confirmation.

Self-check before the visit: choose one shipped lot, one purchased lot, and one subcontracted batch from the same month, and reconstruct their full linkage with the people who did the work, not the quality manager alone. Time the exercise. Every missing link, every record that only the manager can explain, and every explanation that starts with usually is a finding waiting for its auditor.

Sources

  1. Wikipedia: Acceptance sampling —— Overview of acceptance sampling including AQL concepts used in product inspection.
  2. Wikipedia: Factory acceptance test —— Overview of factory acceptance testing performed before shipment or installation.

Frequently asked questions

Is a failed sample during the audit worse than no sample at all?

No, and auditors explicitly say so. A notified sample that fails is a detection event the system should handle through containment and correction. A passed sample from a lot that was secretly sorted before the auditor arrived is an integrity failure that undermines the entire audit. The honest fail with a recorded response builds confidence; the staged pass destroys it.

How do auditors test traceability in practice?

By pulling the receiving records for bought-in parts and matching them against the finished goods traceability for the same period. Gaps usually appear where factories inspect finished goods carefully but accept incoming materials on trust, so a defect found at final test cannot be traced to any lot of components. Auditors follow this direction deliberately because purchasing control is one of the weakest areas in small and mid-size plants.

What should a factory do about outsourced processes before the audit?

Either bring the process inside the documented system with its own controls and records, or apply explicit controls to the subcontractor with evidence. What fails is the middle ground: outsourcing a regulated or quality-critical step with nothing more than a purchase order and a handshake. Auditors treat every outsourced process as part of the system boundary, not outside it.

How can process monitoring records be made audit-proof?

Check that every monitoring device has a valid status, that operators record actual values rather than ticks, and that out-of-tolerance readings trigger the documented reaction. Auditors cross-check process records against equipment logs, so invented entries written at the desk fall apart quickly. One week of genuine records beats three months of backfilled sheets.

Ready to cast for your next campaign?

Tell us the brief — bookings team replies within 48 hours with a curated shortlist from our 480+ roster.

Request a Casting
← Back to Fashion Model BCN