When Bulk Is Wrong: How Defect Claims, Rework and Cost Are Actually Split

There is a reason this subject is discussed in phone calls and not in order confirmations. Both sides have an interest in ambiguity until the moment a problem appears, and by then the ambiguity is expensive for whoever has the weaker record.

The framing that helps is to stop thinking about blame and start thinking about evidence. A defect claim is not settled by who is more reasonable. It is settled by what can be shown: what was approved, what was specified, what was inspected, and what was recorded. Most of the outcomes that surprise brands are the result of a missing record rather than a disputed fact.

What follows is organised by who carries the cost. Three buckets, a remedy decision, and then the clauses that prevent the argument. It is written from the factory side, where we sit, which means it includes the cases where the factory pays.

Faults That Belong to the Factory

These are the cases where the factory carries the cost, and in a well-run relationship they are also the cases where no argument happens, because the evidence is unambiguous.

• The garment differs from the approved sealed sample. Not from the specification — from the physical sample that was signed. If the approved sample has a particular seam finish and the bulk does not, the bulk is wrong regardless of what the paper says.
• The workmanship is outside what was agreed. Puckered seams, uneven hems, visible stitching where there should be none, a closure that does not sit flat. These are execution faults, and the factory's own inspection regime should have caught them before shipment.
• Measurement is outside the stated tolerance at the points that were agreed and measured to the agreed method. This is why the measurement sheet matters: a garment outside tolerance against an agreed point is a clear case; a garment outside tolerance against a point nobody agreed to measure is an argument.
• Materials substituted without approval. A different interlining, a different lining, a closure from a different source. If a substitution was made and not declared, the factory carries the consequence even if the result looks acceptable.
• A process step was skipped or altered. A pressing operation, a stabilising step, a reinforcement that the construction called for. Skipped operations usually show up late, which makes them expensive.
• Damage caused in production or packing. Needle marks, press marks, snags, crushed embellishment. These are factory-side by definition.

Two things are worth noting about that list. The first is that every item is evidenced by a record that a competent factory already keeps — the sealed sample, the measurement sheet, the construction notes. The second is that in each case the factory could have prevented the argument by declaring rather than concealing: a declared substitution is usually a conversation, an undeclared one is always a claim.

Faults That Belong to the Specification

This is the bucket buyers find hardest to accept, because the garment is defective and the fault is theirs. It is also the bucket where a factory that communicated clearly is protected and one that stayed quiet is not.

• The specification produced the fault. A neckline drafted too wide for the size range, a seam allowance too narrow to hold, a construction that cannot work in the chosen fabric. If the factory made the specification faithfully and the result is wrong, the specification is the cause.
• The fabric behaves badly and the fabric was specified. A shrinkage rate that distorts the panel, a colour that shifts when pressed, a weight that will not hold the silhouette. A factory is responsible for telling you what the fabric did in testing; it is not responsible for the fabric being unsuitable if you chose it against that advice.
• An unresolved instruction was resolved by default. Where a tech pack is silent and the factory then makes an assumption, the assumption is the risk — but only if it was not raised. A factory that flagged the gap and got no answer is in a very different position from one that decided quietly.
• The tolerance was never stated. A garment that differs from the spec by an amount nobody defined as acceptable cannot be out of tolerance. This is the most common reason a claim fails on the buyer's side, and it is entirely preventable.
• The approved sample itself was wrong. If the sample was approved with the fault visible, the bulk reproducing the fault is compliant. Approving a sample is the decision that sets the standard, and it is a real decision rather than a formality.

The pattern in that bucket is that the buyer's protection is attention at sampling rather than pressure at delivery. A brand that spends an hour on the sealed sample and the measurement sheet will eliminate most of the situations in which it would otherwise be carrying the cost.

Faults That Are Nobody's Until They Are Somebody's

The awkward middle. The garment was made correctly to a correct specification and is still wrong when it reaches the customer, because something happened between the factory gate and the shop floor.

• Transit damage. Compression creasing, water ingress, handling damage. The question is what the cartons were specified to withstand and who arranged the packing — which is why packing specifications belong in the order rather than in a conversation.
• Condensation and humidity. A container crossing climates can create moisture damage that looks like a manufacturing fault on arrival and is not. Ventilation and desiccant are decisions, and if nobody made them, nobody is accountable until both sides agree who should have.
• Warehouse handling at destination. Damage after receipt is the buyer's, but the boundary is where the goods were accepted. If the receiving inspection was not done at the right moment, the boundary moves, and it moves against whoever did not inspect.
• Fabric batch variation within the agreed standard. Two lots that both pass the agreed test can still look different side by side. If the standard did not cover batch consistency, this is a specification gap rather than a fault.
• Shrinkage that appears only after the customer's first wash. If the care instruction was derived from testing, the responsibility sits with the tested expectation. If it was written from the fibre list, this is a specification failure and it will repeat.

These cases are where a relationship is actually tested, because neither side is clearly at fault and the cost has to land somewhere. The pragmatic approach is to agree the boundary in advance — FOB, delivery, or acceptance after receiving inspection — and then honour it even when the outcome is inconvenient. A boundary that is only applied when it favours one side is not a boundary.

Rework, Credit or Remake

Once the responsibility is established, the remedy is a separate decision with its own economics. Four options, in ascending order of cost.

Two considerations decide between them. The first is whether the fault is random or systematic. A random fault affects a proportion of the order and sorting finds it; a systematic fault is in every piece and sorting is pointless. The second is time — which remedy is available depends on how close the delivery window is, and that often forces a more expensive option rather than a cheaper one.

The remedy that protects the relationship best is usually the one that addresses the process rather than the shipment. A remake that reproduces the fault is worse than a credit, because it proves the correction was never made. Where a fault is systematic, the useful question is not who pays for this order but what changes before the next one.

The Clauses That Prevent the Argument

Every dispute above was decided by a record or its absence. These are the order-level commitments worth having before production starts, most of which cost nothing.

Eight lines, none of which requires a lawyer, and together they turn the question from who pays into which of the agreed options applies. That reframing is worth more than any negotiation held after a problem appears, because it moves the discussion from positions to the document.

The last line is the one most often left out and the one that costs the most when it is needed. A brand that has never discussed remedies will, under time pressure, choose the most expensive one available and resent it. A brand that agreed the options in advance chooses deliberately, and both sides know what is coming.

How the Argument Usually Starts

Claims follow a small number of patterns, and knowing which pattern you are in tells you what the conversation is going to be about before it happens.

1. The comparison problem. The bulk does not look like the sample. Both sides are looking at a different reference — the factory at the written specification, the brand at a garment it approved months ago. The argument is resolved by producing the sealed sample or it is not resolvable.
2. The tolerance problem. The garment is out by an amount that one side considers a fault and the other considers within normal variation, because nobody defined what normal was. This is the most common and the most avoidable.
3. The fabric problem. The fabric behaved differently in production from how it behaved in the sample room, and the question is whether anyone tested for that before cutting.
4. The cumulative problem. No single piece is wrong, but the run as a whole looks inconsistent — slight colour variation, slight placement drift, slight measurement spread. Hard to claim precisely and very damaging commercially.
5. The discovery problem. The fault is found by the customer rather than at inspection, which means the order has already shipped and the timescales for any remedy have compressed.

The fourth is the one to watch. A cumulative quality problem is the hardest to evidence because there is no defective piece to point at, and it is the one that most damages a brand's selling because it becomes visible when garments are seen next to each other rather than individually. It is prevented by consistency controls rather than by inspection, and consistency is a production discipline rather than a quality-control one.

What a Claim Should Include

A claim that arrives as an email saying the order is wrong will take weeks. A claim that arrives as a structured package usually resolves on the facts.

The last line is the one most often missing, and its absence is the main reason claims drag. A buyer who says what they want — sorting, credit, remake, or a change to the process before the next run — gives the factory something to accept or counter. A claim that describes a problem and waits for an offer is a claim that will be met with the smallest offer available.

Two small disciplines make the rest work. Photograph the fault and the sealed sample together in the same frame, in the same light, at the same distance. And count the affected pieces rather than estimating a percentage, because an estimated percentage invites a counter-estimate and a counted figure does not.

Prevention: The Half Hour That Pays

Every category in this article can be substantially reduced by a short set of actions taken before production rather than after. None of them is difficult; all of them compete with the pressure to start cutting.

• Seal a sample and keep one each. Two physical references, signed and dated. This single practice eliminates most disputes in the first two categories.
• Write the tolerances down. Measurement points, methods, and what counts as passing. An undefined tolerance is not a tolerance.
• State the fabric's tested behaviour on the order. Shrinkage, colourfastness, and how it presses. If the fabric has a known weakness, writing it down converts a future claim into a known risk.
• Agree what has to be declared. Specifically: substitutions of materials and components. A declared substitution is a conversation; an undeclared one is a claim.
• Record the machine settings for the decisive operations. Not for everything — for the operations where the sample was right and a second run would be guesswork.
• Agree the risk transfer point and the packing specification. So that transit damage has an owner before it happens.
• Inspect inline, not only at the end. A fault found during production can be corrected; the same fault found at final inspection is a shipment.

Read as a list of costs, all seven are close to free. Read as a list of avoided arguments, they are the cheapest thing on the order. The reason they get skipped is that they do not produce anything visible, which is precisely the category of work that a deadline removes first.

When the Fault Reaches the Customer

A different situation with a different arithmetic. Once garments are with customers, the options narrow and the cost per affected piece rises: returns processing, refunds, replacements, and the less quantifiable cost of the customer's next decision.

• Systematic faults become visible in the returns rate long before anyone connects them to a production cause. A style with a slightly short sleeve or a slightly tight armhole returns at a rate that looks like a sizing preference and is a measurement problem.
• A fault that appears after the first wash is a care-instruction or specification issue and will continue for the life of the style rather than for the affected batch.
• A fault that appears after a few wears is often recovery or construction rather than material, and it is the hardest to evidence because the garment has been used.
• The remedy at this stage is almost always the buyer's unless the fault is clearly systematic and attributable, which is why prevention is worth more here than anywhere else.

The operational point is that returns data belongs in the development conversation. A brand that tracks returns by style and by reason will find construction issues that never appeared at inspection, and it will find them early enough to change the next run. A brand that treats returns as a customer-service metric will keep rediscovering the same fault one season at a time.

The Cost Nobody Puts in the Model

There is a cost attached to how a claim is handled that never appears on either side's arithmetic: what it does to the working relationship, and through that to the next order.

A factory that takes a claim it considers unjustified will price that experience into the next quotation, whether or not it says so. A brand that pursues a claim without a record will find that the factory becomes slower to concede anything else — including the things it would normally absorb as goodwill. Neither of those is retaliation. It is just how a relationship is repriced after a disagreement.

The counterpart is that a factory which concedes a fault it did not cause, in order to keep the peace, has taught the brand that the claim route works without evidence. Both behaviours are common and both are expensive over a few seasons.

The way out is the one this article keeps returning to: make the split a matter of record rather than of negotiation. When both sides know that a comparison against a sealed sample decides the case, a claim stops being a contest of resolve. That is worth more over time than winning any individual argument.

FAQ

Who pays for a defect in bulk?

It follows from the evidence rather than from the size of the complaint. If the bulk differs from the sealed sample, or the workmanship is outside what was agreed, or materials were substituted without approval, the factory carries it. If the specification produced the fault and it was followed faithfully, the specification is the cause.

We never kept a sealed sample. What now?

The dispute becomes about the specification and the photographs, which is a weaker position for both sides. Start keeping one from the next order and it will not recur; retrofitting the practice to this claim will not resolve it quickly.

Is a credit as good as a repair?

It solves the shipment and not the process. A credit on a systematic fault means the same conversation next season. Where the fault is systematic, the useful outcome includes a change to the construction or the process, not only a number.

The goods were damaged in transit. Whose claim is it?

It depends on the Incoterm and on whether insurance covers the loss. The boundary matters more than the answer: agree where risk transfers, then honour it. Where the damage is moisture-related, the packing specification is usually the more useful question than the claim.

Only a small percentage is affected. Is it worth raising?

Yes, and not for the money. A small, systematic percentage in one order is usually a larger percentage in the next one. Raising it early, with the affected pieces identified, is the cheapest form of quality management available.

How long do we have to raise a claim?

Agree that with the order rather than discovering it. Most claims are strongest before the goods are accepted and weaken after they are mixed into stock or shipped onward, and some remedies stop being available at all once the order has left. Put a window in the order terms and inspect at the right moment rather than late.

What if the factory simply refuses the claim?

Then the outcome depends on the record you hold: the sealed sample, the agreed tolerances and the inspection criteria. Those are the documents that make a refusal untenable rather than merely awkward. Without them, both sides are relying on goodwill, and goodwill is not a term of the order.

Working With Luxudress

Luxudress is the factory-direct front end for womenswear production across production facilities in Guangzhou and Dongguan, covering development, sampling, material sourcing, manufacturing, inspection and export packing.

We will seal and sign a sample and ask you to seal one too, put measurement points and tolerances in writing rather than in a conversation, declare any substitution before we make it rather than afterwards, and agree the remedy options with the order rather than after a problem. Where a fault is ours, we will say so; where it comes from the specification, we will say that too, with the record to show it.

If you have a claim in progress, send the sealed sample, the measurement sheet and the affected pieces. You can also see how we run inspection, read how AQL inspection works, or read why approved samples change in bulk.

Request a quote and we will return a costed plan including the inspection and approval terms it assumes.