Rework or Discount? Costing a Defective Underwear Lot (2026)

A container of 5,000 custom OEM boxer briefs arrives with a print registration drift of 1.5 mm on two colourways. The factory offers 8 per cent off, or rework in three weeks. Most buyers pick one by feel. The two options are comparable on one number: the credit rate at which rework stops being worth doing. This page builds that number, then sets the legal boundary around it, because a discount is not always a lawful way to close a defect file.

1. What is the real question when a defective lot lands?

The real question is not “how much off”. It is “which cure costs less once the goods are already in my warehouse”. Three cures exist, and they are not interchangeable:

• Sort — 100 per cent inspection at destination, remove the affected units, ship the rest. Converts an unknown defect rate into a known one.
• Rework — repair at the factory. The goods travel twice more, and the delivery date moves.
• Credit — accept the goods and deduct a percentage from the invoice. The delivery date holds, the defect stays in the market.

Each cure has a cost that lands in a different month, which is why they feel incomparable. Put them on one index and the choice becomes arithmetic. The arithmetic below is deliberately built on an index rather than on anyone’s price list: set the FOB unit price equal to 100 and every other cost as a fraction of it. Your own numbers substitute in directly.

2. How do you calculate the break-even credit rate?

The break-even credit rate is the discount at which the cost of accepting the goods equals the cost of reworking them:

d* = (rework labour + return freight + second outbound leg + duty and tax cost + delay premium) ÷ (affected units × unit price)

The denominator is the money at stake: affected units multiplied by the unit price. The numerator is everything rework actually costs, including the legs buyers leave out. Work the same lot at three different affected shares and a pattern appears:

*Cost index points, unit price = 100. Repair labour is indexed at 6 per unit, which is the order of magnitude for replacing a waistband elastic or re-stitching a leg opening including thread, trim and re-inspection. Freight, duty and delay are fixed legs: they do not shrink when fewer units are affected.

The pattern is the useful part. The break-even rate falls as the affected share rises. At a fifth of the lot, rework only pays if the credit offered is below about 13 per cent. At the whole lot, rework pays below about 7.4 per cent. The reason is structural: three of the four cost legs are fixed, so they amortise across the units being repaired. This is also why factories open with “10 per cent off, no return” on a small affected share — that offer is often above the buyer’s break-even rate, and the buyer has not done the division.

Two adjustments change the answer more than the arithmetic suggests:

• If the credit is applied to the whole invoice rather than to the affected units, the effective rate on the affected units is several times the headline. A “5 per cent off the order” on a lot where 12 per cent of units are affected is a 41 per cent concession on the units in dispute.
• If the delay premium is zero because the goods will sit in your own warehouse anyway, the numerator loses its largest fixed leg and the break-even rate drops sharply. Buyers who never pay for air freight should not price delay as if they might. When the premium does apply, take it from the freight forwarder’s quote as the air-minus-ocean difference per kilo multiplied by the kilos moved, not as a percentage of the order value.

3. What goes into the rework cost that buyers leave out?

Four legs get omitted in almost every first-pass calculation, and they are the ones that decide the outcome:

1. The return leg. Goods going back to the factory leave your country as a return and arrive as an import. Unless the goods are returned under a regime that preserves their status, duty paid on the first import is not automatically refunded, and the second export can attract its own handling.
2. The second outbound leg. You pay freight twice, and on a per-unit basis the second shipment is smaller, so the rate per kilo is worse.
3. The delay premium. If rework costs you the sailing date, the next available option is air freight on part of the order to protect a launch, and that premium belongs in the numerator, not in a separate “logistics” line.
4. The re-inspection. Reworked goods are not inspected goods. A second AQL inspection at the factory, plus a second check at destination if the first one missed the defect, both cost money and both consume days.

The one leg people over-count is labour. Rework on a knitted undergarment is mostly a waistband, a leg opening, a label or a stitch line; if the defect is in the fabric itself — a shade band, a shrinkage difference, a print that has cured wrong — there is no repair, only replacement, and replacement is priced as new production. That distinction decides most cases before the arithmetic starts.

4. Which defects should never be discounted?

Some defects are not a pricing question at all. A discount moves money; it does not move a product into compliance. The boundary is set by mandatory rules, and it is sharp:

Below that line, a credit is legitimate. Above it, a credit is a way of putting a known non-compliant product into a market, and the buyer who accepts it inherits the exposure — which is usually larger than the discount. This is the point at which the cheapest-looking settlement becomes the most expensive decision in the file.

5. What does CISG Article 50 actually allow you to deduct?

Article 50 of the CISG gives the buyer a right to reduce the price, and it defines the size of the reduction in a way that most email negotiations ignore: the reduction must be in the same proportion as the value that the goods actually delivered had at the time of delivery bears to the value that conforming goods would have had at that time.

That is a valuation test, not a negotiation. If goods worth 100 in conforming condition are worth 82 as delivered, the reduction is 18 per cent — and the number has to be defensible, because it is derived from relative value at the time of delivery, not from the inconvenience the buyer feels. In practice the delivered value is taken at its realisable price in the condition received: the secondary-market or jobber price for the sorted goods, a documented sell-through loss against the plan, or the buyer’s own resale deduction on the affected units. Whatever basis you use, put it in the file — the percentage is the output of that comparison, not an input to it. Three consequences follow:

• “20 per cent off” needs a valuation behind it. A buyer who asks for a round number without a value comparison is asking for a favour, and a favour is what the seller will treat it as.
• Cost of replacement is not the measure. Article 50 measures the reduction against value, not against your cost of buying elsewhere. Cost-based recovery travels under the damages articles (Article 45(1)(b) and Articles 74 to 77), which require the loss to have been foreseeable at the time of contracting and require the buyer to mitigate.
• The reduction applies to the non-conforming part. Article 51(1) extends Articles 46 to 50 to the part of the goods that does not conform, which is why the affected share has to be expressed in units rather than in cartons.

6. Why can rework sometimes forfeit the credit?

Because the two remedies are sequenced, not stacked. The second sentence of Article 50 bars the buyer from reducing the price if the seller has remedied the failure under Article 37 or Article 48, or if the buyer refused the seller’s offer to do so. Article 48(1) lets the seller remedy at its own expense even after the delivery date, provided it can do so without unreasonable delay and without causing unreasonable inconvenience or uncertainty of reimbursement — and the buyer keeps the right to damages for the delay.

In practice this creates a fork with a deadline:

Article 46(3) sets the same timing for repair: a request for repair must be made either together with the notice under Article 39 or within a reasonable time afterwards. A buyer who first writes “we will sort it ourselves and deduct” and only later asks for rework has usually spent the option.

7. Which defects suit rework, and which suit a credit?

Sort by two properties: is the defect repairable, and is it uniform across the lot.

The pattern in the table is that repair fits scattered defects and credit fits uniform ones. A uniform defect means every unit in the affected group is worth less by roughly the same amount, which is exactly the situation Article 50’s proportional reduction describes. A scattered defect means most units are fine, so a percentage credit overpays for the ones that are not. Clustering decides the quantity to settle; repairability decides the cure.

8. When is a 100 per cent sort cheaper than either?

Sorting is the cure buyers forget because it does not fit either negotiation script. It is usually the right first move when the defect rate is genuinely unknown, because it converts the estimate into a count, and a count is what makes both the credit and the rework claim defensible. It also protects the Article 50 valuation: you cannot value delivered goods against conforming goods without knowing how many units are in each state.

9. How do you estimate the affected share from an AQL sample?

With ISO 2859-1, at the inspection level and AQL written into the purchase order. The relevant acceptance numbers for single sampling at general inspection level II, AQL 2.5, are these:

Read the last column carefully: at an AQL of 2.5 the acceptance number still tolerates 4.4 to 5.6 per cent defectives in the sample, so the table is a ceiling on the defect share, not a certificate of zero. A lot that passed inspection is not a lot without defects, and the credit conversation frequently starts from a passing report. Two rules follow:

• The sample gives an estimate, not a count. Take the number of defectives in the sample, divide by the sample size, and treat the result as the working estimate for the lot — then true it up against the sort tally. Article 50’s valuation needs the count, so the estimate is a placeholder, not a settlement.
• Critical defects are not sampled the same way. A critical defect class is normally specified at zero acceptance regardless of AQL, and a lot with one critical defect is a different file from a lot with ten cosmetic ones.

If the report itself is unclear on defect classification, the reading method matters more than the arithmetic — the classification drives everything downstream. See how to read an underwear inspection report for the field-by-field method.

10. What has to be in the purchase order before any of this happens?

Every lever described above is written before production or not at all. After the container arrives, the only remaining lever is the balance payment, and that is a blunt instrument. The clauses that make the arithmetic work:

• Defect definitions by class, with examples, so the inspection report and the claim use the same vocabulary.
• The AQL and inspection level, stated as ISO 2859-1 with the level and the AQL value, not as “standard inspection”.
• A cure hierarchy in order of preference — sort at destination, then rework, then credit — with a stated deadline for the seller to elect a cure.
• A valuation method for price reduction, expressed as relative value at delivery, so Article 50 is operational rather than theoretical.
• The cost legs that rework must carry: return freight, second outbound leg, duty and tax cost, re-inspection, and the delay premium where a sailing date is missed.
• The balance payment trigger, tied to the agreed cure being evidenced rather than to the bill of lading alone. The clause wording that works is set out in the supply agreement clauses page.

For a first order, these clauses are also the cheapest test of the counterparty. A supplier who will not accept a stated cure hierarchy and a valuation method is telling you what the negotiation will look like later; the same set of questions, asked before the order, is what the twelve-point supplier verification checklist is for.

11. What does this look like on our side?

Unitex is a sourcing and trading company: we place custom OEM underwear production with partner factories that hold amfori BSCI and OEKO-TEX Standard 100, and we handle the specification, the inspection and the claim file. In a rework-or-credit discussion we provide both numbers rather than one: the repair cost per unit with the fixed legs itemised, and the value comparison behind any proposed credit rate, in writing, within the notice window. For licensed-category underwear the minimum is 3,000 pieces per style per colour, and bulk production runs 45 to 60 days on a new order, which is why the cure decision usually has to be made in the week the container lands rather than after the season is priced. If you want to see how the arithmetic works on a live order, send the inspection report and the invoice value and we will run both columns.

Next step: before the next container is opened, fix the defect classes, the AQL, the cure hierarchy and the valuation method in the purchase order. Then read the inspection report against that document, not against the supplier’s summary of it. The three pages that carry the rest of this file are the evidence standard for a quality claim, the notice windows and replacement timeline, and the first-order failure modes that make the first container the expensive one.

Published: 2026-10-31 · Xiamen Unitex Trade Co., Ltd. — OEM & ODM underwear sourcing partner

Unitex Underwear (Xiamen Unitex Trade Co., Ltd.) — OEM/ODM underwear sourcing and supply-chain partner in Xiamen, China; in the trade since 2009, incorporated in 2013. Based in the Xiamen–Quanzhou–Jinjiang knitwear cluster, Fujian Province, south-east China.