At 09:10, the buyer says the shipment is unusable. At 11:30, the supplier offers replacement parts. By 15:00, the buyer's sales team says replacement will arrive too late. At 18:20, finance asks whether the remaining invoice can be withheld. The next morning, someone says “send a legal demand.”

That sequence is normal. What is dangerous is treating every step as a choice between accept everything and start a lawsuit.

A quality dispute has several resolution paths. The best one depends on what is wrong, whether the defect can be verified, how quickly the goods are needed, whether cure is realistic, how much value remains in the relationship, and what the governing contract and law actually allow.

First decision node: is the problem measurable?

Before choosing a remedy, convert the complaint into a testable statement.

Weak statement: “The finish is terrible.”
Better statement: “The approved reference is gloss 15–20 GU at the stated measurement geometry; sampled units measured 31–38 GU under the agreed method.”

Weak statement: “The parts do not fit.”
Better statement: “Hole spacing on drawing Rev C is 120.0 ±0.5 mm; the documented sample measures 122.1–122.8 mm.”

If the parties cannot agree what is being measured, arguing about remedy is premature.

Path 1: local repair or rework

When it works

Rework is attractive when the defect is technically reversible, the affected population can be identified and a competent local provider can correct it faster than replacement.

What people get wrong

Wrong: “Rework is cheaper, so it is obviously the answer.”
Better: price rework, inspection after rework, downtime, scrap risk, warranty responsibility and who signs off the corrected goods.

A repair can create a new quality dispute if nobody agrees on the acceptance standard after the work.

Decision questions

  • Can the defect be corrected without reducing safety, performance or expected life?
  • Who owns the rework method?
  • Who verifies completion?
  • Does rework affect certification or warranty?
  • Is there enough time before the goods lose their commercial purpose?

Path 2: replacement

When it works

Replacement is useful when the defect is clear, production can restart quickly and the buyer still has time to receive conforming goods.

What people get wrong

Wrong: “Replacement makes the buyer whole.”
Better: distinguish replacement of goods from the costs created by the delay.

New units do not automatically solve expedited freight, lost launch windows, sorting labor or downstream claims. Whether those losses are recoverable is a separate legal question.

Under the CISG, if it applies, remedies depend on the character of the non-conformity and the Convention's requirements. Do not use a general statement such as “international law gives the buyer replacement” without checking scope, seriousness, notice and the actual contract.

Path 3: supplier-led cure

Cure is broader than replacement. It can involve missing documents, corrected packaging, software update, parts, technician attendance or another step that fixes performance.

UNIDROIT Principles Article 7.1.4 provides a structured cure concept: notice, timing, appropriateness and the aggrieved party's legitimate interest matter. The Principles are soft law, however; they do not create an automatic worldwide right for a supplier to enter the buyer's facility and fix goods.

What people get wrong

Wrong: “The supplier offered to fix it, so the buyer must accept.”
Better: ask whether the proposed cure is legally available, timely, technically sufficient and commercially useful.

A cure after the buyer's critical sales date may be technically perfect and commercially worthless.

Path 4: price adjustment or credit

This is often the fastest path when the goods retain economic value.

The key question is not “What discount feels fair?” It is “What measurable loss of value or corrective cost does the discount represent?”

Possible anchors include:

  • verified local rework cost;
  • reduced resale value;
  • missing feature value;
  • extra inspection/sorting cost;
  • agreed percentage tied to affected units.

What people get wrong

Wrong: negotiate a round number first and explain it later.
Better: build a one-page calculation both finance teams can reproduce.

Price adjustment is especially useful where the parties want to keep supplying each other and the defect does not create a safety or regulatory problem.

Path 5: controlled acceptance with reservation

Sometimes the buyer must keep production moving even while a claim remains open.

It may be possible, depending on governing law and contract, to accept/use some goods while expressly preserving disputed rights. But the legal effect of acceptance, use, resale and reservation varies. Do not assume an email saying “accepted under protest” has the same consequence everywhere.

Operational guardrail

If goods must be used, record:

  • which units were used;
  • why use was commercially necessary;
  • what condition was observed first;
  • what rights the buyer says it is preserving;
  • what the supplier says about that reservation.

Local counsel should confirm the effect before the action becomes irreversible.

Path 6: return or rejection

Rejection sounds decisive, but it creates physical questions immediately:

  • who stores the goods;
  • who pays return freight;
  • whether export/import documents need correction;
  • whether goods may be altered before return;
  • who bears deterioration risk while they wait;
  • whether the buyer has a duty to preserve them.

What people get wrong

Wrong: “We rejected them, so they are the seller's problem now.”
Better: separate legal rejection from physical custody and evidence preservation.

The UK Sale of Goods Act offers one national example of rules around quality, sample and remedies, but its terminology and effect should not be transplanted into a contract governed elsewhere.

Path 7: split the shipment

Not every quality problem affects the whole lot.

A commercially smart solution may be:

  • release conforming units;
  • quarantine borderline units;
  • reject a defined subset;
  • rework another subset.

This reduces the value at risk and makes later negotiation more precise.

What people get wrong

Wrong: one failed sample equals automatic rejection of every unit.
Better: determine what the agreed sampling plan, defect classification and technical evidence actually support.

Path 8: structured commercial negotiation

Negotiation works best after each side knows its BATNA—the realistic alternative if no deal is reached.

A useful settlement sheet has four columns:

Issue Buyer position Seller position Verifiable middle ground
affected quantity 100% 8% third-party sample
repair cost $X $Y two local quotes
delivery impact six weeks two weeks updated production schedule
credit 25% 5% affected-unit model

Now the conversation is about variables rather than insults.

Path 9: mediation

Mediation may suit a relationship worth saving, a fact pattern where both sides face uncertainty, or a dispute with many commercial issues that a court remedy would not elegantly solve.

But mediation is not magic. If one side needs an urgent injunction, binding technical determination or enforceable order, another route may be necessary.

Check the dispute clause: some contracts require negotiation or mediation before arbitration/litigation; others do not.

Path 10: expert determination or technical neutral

For a dispute centered on a technical specification, the parties may agree to use a neutral expert to decide a defined question.

Examples:

  • whether coating thickness satisfies the named specification;
  • whether failure mode is consistent with packaging damage;
  • what proportion of sampled units falls outside tolerance.

Do not ask the expert to decide vague legal questions unless the contract clearly creates that authority.

Path 11: arbitration

Arbitration may be appropriate where the contract contains a valid arbitration agreement, the value justifies the process and cross-border enforceability matters.

ICC's 2026 Arbitration Rules took effect on 1 June 2026, but they matter only if the parties' arbitration agreement and applicable framework make ICC arbitration relevant. Other institutions and ad hoc procedures exist.

The New York Convention provides an international framework for recognition and enforcement of qualifying arbitral awards, subject to its scope, local procedure and defenses. It is not an automatic global collection button.

Path 12: litigation

Court proceedings can be necessary for binding relief, urgent measures, third-party issues or where the contract points there.

Before filing, ask:

  • which court has jurisdiction;
  • governing law;
  • service requirements;
  • expert evidence rules;
  • preservation/disclosure obligations;
  • limitation period;
  • where the defendant has assets;
  • how a judgment would be recognized or enforced elsewhere if necessary.

Winning in the wrong forum is an expensive way to learn about enforcement.

A practical route-selection matrix

Situation First route to test Reason
clear cosmetic defect, goods usable credit/rework preserves value
technical cause disputed joint inspection/expert fact comes before remedy
urgent production need replacement/local cure time dominates
safety/regulatory concern quarantine + specialist advice ordinary compromise may be unsafe
relationship valuable, facts mixed structured negotiation/mediation flexible package possible
binding technical issue expert route if contract allows narrow decision
high-value cross-border breach arbitration/litigation analysis enforceability matters

A useful discipline is to set an escalation trigger when the parties choose an informal path. For example: “If the joint inspection is not completed by Friday, counsel will issue the contractual notice,” or “If the rework trial fails two agreed acceptance checks, replacement becomes the default commercial proposal.” This stops negotiation from turning into indefinite drift.

Also record who has authority to close each path. A factory quality engineer may approve a rework method but not a $75,000 credit. A buyer's procurement manager may negotiate a discount but not waive a claim. Resolution stalls when the technical people think they have a deal that the financial or legal decision-makers cannot approve.

Finally, price the switching cost between paths. Moving from rework to replacement after two failed weeks is more expensive than deciding the fallback before the first rework starts. A staged plan should therefore name the fallback in advance, including what evidence will trigger it.

The sequence matters more than the label

The best resolution process often changes over time.

Morning: preserve evidence.
Day 2: joint inspection.
Day 4: rework proposal.
Day 7: quantified credit discussion.
Day 12: lawyer-to-lawyer demand.
Day 20: mediation or formal filing if the commercial path fails.

That is not indecision. It is staged, controlled escalation with an explicit fallback.

Three comparisons worth remembering

Wrong: pick a remedy before proving the defect.
Better: prove enough to make the remedy proportionate.

Wrong: treat “supplier will fix it” as the end of the dispute.
Better: test whether cure protects timing, quality, certification and downstream commitments.

Wrong: threaten arbitration because it sounds serious.
Better: map forum, cost, evidence and enforcement before escalating.

Bottom line

A nonconforming-goods dispute is not one road with litigation at the end. Repair, replacement, cure, credit, controlled acceptance, partial release, negotiation, mediation, expert determination, arbitration and court can each be rational in the right sequence.

The route should follow the verified defect, commercial deadline, relationship value and legally available remedies—not emotion or the loudest email in the chain.

General commercial information only, not legal advice. The availability and effect of rejection, cure, replacement, damages, mediation, arbitration and court remedies depend on contract terms, governing law and jurisdiction.

Related Reading

Sources and Scope Notes