When to Reject a Luxury Handbag Shipment

The decision nobody wants to make

A refusal is not an accusation. It is a routing decision inside a quality system, and the system is supposed to produce it sometimes. If a programme never refuses anything, either the specification is too loose or the inspection is not actually looking. The emotional difficulty comes from treating refusal as a personal failure rather than as the fourth possible outcome of a gate that has three others. A programme that has never refused a lot is usually a programme whose specification is too loose to fail anything, and that is the more dangerous position.

The practical difficulty is different: refusal costs time the season may not have. That is the real calculation, and it is why the decision needs a rule rather than a mood. A written threshold, agreed before the order, converts an argument into a lookup.

Case one: the sporadic fault

A maison ordered 500 pieces of a structured calfskin shoulder bag in a deep navy. At final inspection, 50 pieces were drawn at general inspection level II. Four pieces carried a misaligned flap edge, measuring 1.2 mm off the reference at the point where the flap meets the body — visible, but only at one corner, and only on four of fifty. The count exceeded the AQL 2.5 threshold of three, so the lot failed the letter of the plan.

The correct answer here was not refusal. The fault was sporadic rather than systemic: it appeared on a small, scattered subset, and when we traced the pieces back through the cutting record they came from three hides at one end of the lot. Sorting the full 500 against a 0.5 mm gauge took a day, separated 31 pieces, and the remaining 469 shipped on schedule. The four-in-fifty finding was a signal to sort, not a signal to rebuild.

Case two: the systemic fault

A second programme, a frame bag in printed calfskin with a plated clasp, reached final inspection with a different pattern. Twelve of the fifty drawn pieces showed clouding at the clasp edge, and every single one came from the same hardware batch. This was not a distribution of faults across a lot; it was one cause producing the same fault repeatedly, and sorting would not have helped, because the affected pieces could not be identified without disassembling each one.

The lot was refused. The hardware batch was quarantined, the plating bath records were reviewed, and the replacement batch was put through 24 hours of neutral salt spray before a single clasp was set. The delay was eleven working days. Releasing the lot instead would have put several hundred bags with failing clasps onto boutique floors, where the failure would have surfaced in the clients' hands rather than at our bench.

Case three: the unexplained fault

The hardest case is neither sporadic nor obviously systemic. A brand received an inspection report showing six minor findings and one major: a lining seam that had opened on a single piece. The count was well inside tolerance, and by the numbers the lot conformed. But nobody could explain why a lining seam had opened, and an unexplained fault is a fault that will happen again.

We held the lot anyway, on the grounds that the cause was unknown rather than the count being high. The investigation took two days and found a tension setting on one machine that had drifted across a shift, affecting seams sewn in a four-hour window. Twelve pieces were identified, re-sewn and re-inspected. Had the lot shipped on the strength of the numbers alone, a seam would have opened on a client's bag and nobody would have known which twelve to look at.

The three conditions that justify refusal

Out of those cases comes a rule with three conditions, and any one of them is sufficient. Refuse when the fault is systemic, meaning one cause producing the same fault across a large share of the lot. Refuse when any critical finding appears, because critical defects carry zero tolerance at any count. Refuse when the count is inside tolerance but the cause is unknown.

• Systemic: one identifiable cause, repeated across the lot, not separable by sorting.
• Critical: any finding that makes a piece unsafe, illegal or unsaleable, regardless of count.
• Unexplained: the numbers pass but nobody can say why the fault occurred.
• Recurring: the same fault appeared on the previous lot and the corrective note has not taken effect.
• Unstable correction: a rework that cannot be verified will simply produce a second, larger problem.

Reading the inspection report correctly

Most disposition arguments come from misreading the report. The accept number is the maximum count that still passes; the reject number is one above it, and at AQL 2.5 on 50 pieces that means three passes and four refuses. There is no middle category. The report also separates major from minor findings, and a lot can fail on majors while passing comfortably on minors, or the reverse. Read the classification before the total.

What refusal actually costs

The cost of refusal is time, and on a 500-piece structured bag programme that time runs from ten to fifteen working days for a re-run: new material released, the line rescheduled, and a fresh final inspection. Against that, reworking one closure on a piece that has already reached a client costs around USD 4.80 in parts and bench time, before return freight, client service handling and the damage to a relationship with a buyer who placed the piece on the floor.

The comparison is not close. Ten working days of delay is recoverable inside a season; a wave of returns across several boutiques is not, and neither is the conversation with a buyer who has lost confidence in the brand's quality. The brands that refuse early and refuse on evidence ship fewer problematic pieces and, over a few seasons, spend far less on remediation than the brands that negotiate every borderline lot into a container.

How to refuse without damaging the relationship

Refusal handled badly damages a programme more than the fault itself. The three things that keep it constructive are evidence, a named cause and a proposal. Evidence means photographs with a scale reference and the measurement that produced the finding. A named cause means an explanation, even a provisional one. A proposal means a route back: sort by Thursday, rework eleven pieces, or re-run with a replacement batch arriving on a stated date.

1. Hold and tag the lot before anything is decided, so no piece moves while the argument runs.
2. Issue the report with photographs and measurements, classified by severity, within one working day.
3. State the disposition you recommend and the reason, rather than asking the other side to choose.
4. Name the delay in working days and the date the corrected lot can be re-presented.
5. Write the root cause note before the next lot starts, and attach it to the specification.

What happens after the refusal

A refused lot is re-presented, and the re-presentation is inspected at the same level as the original draw, never at a reduced one. The programme also moves to tightened inspection under the switching rules of ISO 2859-1 after two refusals inside five consecutive lots, which lowers the accept number until five consecutive lots pass. That is the mechanism by which the system protects itself, and it is worth explaining to a brand in advance so the tightening is not read as a punishment.

How to avoid reaching this decision

Every case above was preventable upstream, and all three were prevented by the same two habits. The first is a first piece sign-off that is genuinely examined rather than waved through, which would have caught the flap alignment in case one before fifty bags existed. The second is intake testing on bought-in components, which would have caught the plating in case two before a single clasp was set. The third case, the drifting machine tension, is caught by inline checkpoints every two working hours rather than by any inspection at the end.

Standing AQL, major defects

2.5, ISO 2859-1, single sampling, normal inspection

Accept and reject at 50 pieces

3 accepted, 4 refused for major defects

Critical defects

Zero tolerance, one occurrence refuses the lot regardless of count

Sort cost

1 to 3 working days on a 500-piece lot

Rework cost

3 to 6 working days, plus full re-inspection at the same level

Refusal and re-run

10 to 15 working days on a structured bag programme

Hold for investigation

2 to 5 working days while the cause is established

Rework cost per piece

Around USD 4.80 in parts and bench time for one closure

Tightened inspection trigger

Two refusals inside five consecutive lots

Return to normal

Five consecutive passing lots under tightened rules

Minimum order quantity

500 pieces per style per colourway

Sampling lead time

6 to 10 working days, 12 to 15 for complex constructions

Bulk lead time

35 to 50 days from approved sample

Shipment terms

FOB Xiamen

Frequently Asked Questions

People Also Ask