The shipment failed inspection: accept, discount, or reject?

Short answer: reject when the defects are functional and the calendar can absorb a re-make, because the factory reworks at their cost and you get a clean lot. Take a discount when the defect is cosmetic, shallow, and cheap to sort — and only if the discount beats your actual remediation cost, which the arithmetic below will show. Accept as-is, no concession, almost never. The one move merchants miss is the fourth option: accept the lot but force a 100% sort before it ships.

This article is the decision layer. How the failure got measured — the sample size, the accept/reject counters, the defect classes on the report — is the math covered in the AQL sampling guide, and this picks up the moment the report says fail.

What "failed" actually means on the report

A pre-shipment inspection fails when the sample crosses a reject number: say 6 or more majors in an 80-unit sample on the standard 2.5 setting. The report gives you three things worth reading closely before deciding anything:

• The class of the failure. A lot that failed on minors is a different decision than a lot that failed on majors, and a single critical defect — safety or regulatory — ends the discussion entirely.
• The defect list. Not just counts: which defects, at which rates, on which units. "Color mis-registration on 4 of 80" is negotiable; "weld cracking on 7 of 80" is not.
• The remedy line. Reports from firms like QIMA typically state whether the lot can be reworked and re-inspected before shipment (QIMA, accessed November 2026). If the factory can re-sort on their floor this week, "reject" costs you days, not weeks.

The decisive question hiding inside all three: what does this defect rate cost you after receipt? That cost depends on your stock position, and it's the variable most merchants leave out of the conversation entirely.

The decision table

Run the failure through these rows in order; the first row that matches usually settles it.

Two things the table is NOT doing: it's not a substitute for the arithmetic (next section), and it's not leverage-neutral — a factory that knows you're weeks from zero will smell it. Which is exactly why the math comes first: you need to know what you can actually afford before the phone call.

Worked example: 3,000 enamel pins, 12 majors in the sample

Illustrative numbers, full arithmetic. Your PO: 3,000 × PIN-EN-043 at USD 0.42 = a $1,260 estimated invoice. The inspection (Level II, 125-unit sample) finds 12 majors — color mis-registration and missing rubber clutches — against a reject number of 8. The lot fails: 12 > 8, a defect share of 12 ÷ 125 = 9.6% of the sample.

PIN-EN-043 sells 11 units/day (illustrative) and you hold 190, which is 190 ÷ 11 = 17 days of cover. Run the three options:

Reject and re-make. The defect is the factory's, so the re-make runs at their cost: assume 15 days production (illustrative) plus a re-inspection man-day at roughly $300 estimated. Your cover is 17 days, so the calendar absorbs it with 2 days to spare. Missed revenue: about zero. Total cost to you: ~$300 estimated in re-inspection and two weeks of waiting. On a $1,260 PO, this wins almost every time.

Accept with a discount. The defect can't be sold retail — mis-registered pins are landfill — so accepting means sorting and writing off. Sorting 3,000 pins at ~150 units/hour and a $18/hour (illustrative) helper: 20 hours = $360 estimated labor. The write-off: 9.6% × 3,000 = 288 units × $0.42 = $121 estimated. Add refund and review exposure on whatever the sort misses. The discount needs to beat ~$500 estimated, which is 40% of the invoice — and a factory facing a $1,260 PO will not concede 40%. This is the structural trap with cheap-unit goods: the invoice is too small for the discount lever to carry real weight.

Accept and sort locally, no discount fight. Same $360 estimated labor and $121 estimated write-off, plus you keep the calendar — but you also absorb every miss. Only defensible if the factory refuses rework AND your cover is under a week.

The verdict: reject, and the pattern generalizes — when unit cost is low, the discount lever is weak and rejection is cheap. The ratio inverts as unit cost rises.

When the discount is real money

Same failure mode, different economics. Hiking socks: 2,400 × HIKE-SK-MRN-L01 at USD 6.80 = a $16,320 estimated invoice. The inspection fails on minors — loose threads, slight needle lines, nothing functional. Here a 10% discount is $1,632 estimated, which comfortably exceeds sorting labor and the write-off on a cosmetic defect you can move as bundle fillers or second-quality sales. Now accept-with-discount is the rational move, and the negotiation writes itself: the report shows the defect classes, your remediation quote shows the cost, and the discount just has to clear a number you computed before the call.

The email that works is short: the failed report attached, the remediation cost stated, one of three closes — re-make at your cost, or a discount of $X, or we reject under the PO terms. Everything in it traces to a document the factory signed (the PO's inspection clause) or a standard they recognize (the sampling plan). Nothing in it traces to feeling.

Risk, terms, and the point where this article ends

Two fine-print items shape what "reject" even means. First, risk transfer: under FOB terms, risk passes to you when goods go on board at the origin port, so a problem discovered after sailing is a negotiation rather than a refusal — the transfer points are defined in the ICC's Incoterms 2020 rules (accessed November 2026). Second, governing law: most cross-border supply contracts fall under the UN Convention on Contracts for the International Sale of Goods, which requires the buyer to notify the seller of non-conformity within a reasonable time (United Nations CISG, accessed November 2026) — one more reason the inspection happens before shipment and the notice goes out the day it fails.

And then the boundary: everything above happens before you accept the goods. Once defective stock has been received and paid for, you're in a different decision tree — credits, return freight, write-offs — covered in the return-to-vendor guide.

One cost worth naming before you choose: while the lot waits for re-make, the SKU keeps selling, and if it hits zero mid-wait the damage isn't theoretical. While the shelf is dark, demand leaks: some buyers substitute, some wait, and a steady share simply buy the same item from a competitor whose listing sits one scroll away. What a dark SKU actually costs you per day is worth computing in advance, and the out-of-stock cost guide walks that calculation.

Common questions

Should I accept a shipment that failed inspection? Only with a concession. Reject outright if the defects are functional or critical and your stock cover can absorb a re-make; accept with a discount only when the discount exceeds your full remediation cost — sorting labor, write-offs, refund exposure. Accepting a failed lot as-is converts a factory's defect into your customer's first impression.

Who pays when a pre-shipment inspection fails? The factory, in the standard arrangement: the defect is theirs, so rework, re-sorting, or re-making happens at their cost, and re-inspection is negotiated onto them. Your costs are the calendar days the fix consumes and the re-inspection fee if you carry it — which is why the inspection clause in your PO matters more than the inspection fee itself.

What's a reasonable discount to ask for on a defective shipment? One that exceeds your actual remediation cost, computed first: sorting labor, the write-off value of unsellable units, and refund exposure. On cosmetic defects with expensive units (10% of a $16,000 estimated invoice is $1,600 estimated) that math works. On cheap-unit goods it usually doesn't — the whole invoice is smaller than your remediation, and rejection costs less.

Can I refuse the goods after they've already shipped? It depends on your shipping terms. Under FOB, risk passed to you when the goods went on board at origin, so refusal at arrival is a claim negotiation, not a refusal. Under EXW or terms where risk transfers later, you have more standing. The transfer points live in the Incoterms 2020 rules, and your PO should name the term explicitly.

Is it better to reject or negotiate? Reject when the defect is systemic and the factory can re-make at their cost within your stock cover — that's the cleanest outcome. Negotiate when the calendar is the binding constraint: a near-empty SKU in peak week makes an imperfect lot with a discount the least-bad option. Compute both costs before the call, then let the numbers pick.

Next step

Before your next inspection, write down two numbers for the SKU in question: days of cover on hand, and the daily revenue the SKU carries. When a report says fail, those two numbers pick the option faster than instinct will — and if the answer is "reject but the cover is short," the out-of-stock cost guide shows what each waiting day is worth, and the return-to-vendor guide covers the aftermath of lots you've already accepted.