Quality Control

When a Pre-Shipment AQL Fails: How to Respec Custom Metal Promo Orders

The scenario: PSI fails with 12 days left to vessel cut-off

A distributor is trying to release a 20,000-piece retail program: 8,000 soft enamel pins, 7,000 zinc alloy keychains, and 5,000 die-struck challenge coins. Goods are packed, the ex-factory balance is pending, and the booking window is tight. A third-party pre-shipment inspection at General Inspection Level II returns a fail. On the applicable sample size, the report records 7 major defects against an acceptance number of 5 and 11 minor defects against an acceptance number of 10. Major findings include front-face plating scratches visible at normal retail distance, jump rings opening under light pull, and enamel underfill around text strokes below 0.30 mm. Minor findings include lot-to-lot color shift, backing cards inserted off-center by more than 2.0 mm, and scattered backside haze.

This is where time gets burned if the original PO only says “good quality,” “no scratches,” or “match artwork.” Those are commercial expectations, not inspection criteria. When PSI fails, the root problem is often not only process execution but also an under-specified product standard: the factory, the inspector, and the buyer are each using a different threshold.

The fastest recovery is not “sort and improve.” It is a written respec issued within 24 hours of the failed report. That respec should define defect classes, appearance zones, measurable tolerances, test methods, re-inspection AQL, quantity disposition, and release gates. With 12 days left to vessel cut-off, every vague phrase costs a day.

1. Freeze release and rank defects by business risk

First, stop shipment release. Do not move cartons to forwarder staging while quality disposition is open. Quarantine stock by SKU, plating batch, production date, assembly line, and carton range if traceability exists. Mixed promo orders often contain one salvageable SKU and one remake-only SKU; treating the order as one undifferentiated problem usually wastes 48 to 72 hours.

For custom metal promo products, sort findings into four decision buckets: safety, function, visual brand impact, and pack-out accuracy. Sharp burrs more than 0.05 mm proud of the edge, plating flakes that can cut skin, or exposed pin posts are safety issues. Jump rings opening below minimum pull force, butterfly clutches slipping off, or warped pins rocking more than 1.0 mm on a flat plate are functional issues. Front-face scratches, unreadable text, enamel voids in logo zones, or visible color drift between adjacent retail units are brand-impacting. Wrong counts, wrong backing cards, inverted barcodes, crushed hang holes, or mixed SKUs are pack-out failures.

A practical triage rule works well in promo merchandising: if an end user can see the defect within three seconds at 30 cm under 800 to 1000 lux, treat it as visual major unless it sits in a defined non-appearance zone. If the item can fail during one event cycle, such as a keychain separating from its ring or a pin twisting because the post is offset by more than plus or minus 1.0 mm, treat it as functional major. This ranking determines where labor goes first. Reworking backside haze while weak jump rings remain in lot stock is the wrong sequence.

• Stop booking release until a revised inspection passes or a written partial-release plan is signed.
• Quarantine by SKU, lot date, plating batch, line, and carton number range.
• Classify every finding as critical, major, or minor with photo references and sample IDs.
• Define front-face zero-defect zones, controlled-cosmetic zones, and non-appearance zones.
• Set the re-inspection date, tightened AQL, and release trigger before authorizing rework.

2. Convert the failed report into a measurable respec

The respec must be tighter than the original PO and usable by both a line supervisor and a third-party inspector without interpretation. Replace subjective wording with measurable criteria: where the defect matters, how large it can be, what lighting applies, what gauge confirms it, and whether it is critical, major, or minor. A good respec is not longer because it is wordier; it is better because it is inspectable.

For custom metal promo products, appearance tolerances can be stated precisely. Examples: no exposed base metal on any front-facing surface; no front-face scratch over 0.50 mm visible at 30 cm under 800 to 1000 lux; backside hairlines acceptable only outside logo zone if each mark is under 2.0 mm and no more than 3 marks occur within a 25 mm diameter area; enamel sink in logo or text zones not to exceed 0.10 to 0.15 mm measured by depth gauge or approved comparator; isolated pits not to exceed 0.30 mm diameter on front face; pin-post center offset not to exceed plus or minus 1.0 mm from approved drawing; backing card placement tolerance plus or minus 2.0 mm for cards up to 90 mm wide and plus or minus 3.0 mm above 90 mm; jump ring residual gap after closure not to exceed 0.15 to 0.20 mm.

This is also the point to tighten the inspection gate. Many promo orders start at General Level II with AQL 2.5 for major and 4.0 for minor. After a first PSI fails on function or retail-facing appearance, buyers commonly tighten corrected defect categories to AQL 1.0 or 1.5 for major while leaving secondary cosmetic points at 2.5. For high-visibility launch items, some buyers also move the re-inspection to General Level III for the remade lot only, especially when the first fail involved hardware integrity or multi-SKU mixing.

Issue found at PSI	Weak wording that creates dispute	Respec wording that can be inspected
Front-face plating scratches	No scratches allowed	No exposed base metal on front face; any front scratch over 0.50 mm visible at 30 cm under 800-1000 lux is major; backside hairlines under 2.0 mm acceptable only outside logo zone
Enamel underfill or pits	Colors should be smooth	Enamel sink in logo/text zone max 0.15 mm; no pit over 0.30 mm dia on front; low spots outside logo zone max 0.20 mm
Loose keychain hardware	Hardware must be strong	Jump ring fully closed, gap max 0.20 mm; split ring pull-force minimum 8 kgf, target 10 kgf; no separation after 10 manual open-close cycles
Color mismatch across lots	Match Pantone as close as possible	Compare to sealed sample under D65 light; no obvious lot shift on adjacent packed units; delta accepted by signed production standard only
Card placement off-center	Packing should be neat	Backing card offset max plus or minus 2.0 mm; hang hole must remain centered and usable; barcode orientation must match approved pack-out photo

3. Choose the right recovery path: rework, partial remake, or concession

Not every PSI failure deserves the same corrective path. Replating finished pins or coins after posts, clutches, solder points, or edge details are complete is often high-risk because stripping can soften relief, reduce antique contrast, enlarge pits, and increase weld or solder failures. Replacing weak jump rings, changing split rings, re-carding product, or sorting cosmetic defects by appearance zone is usually practical and fast. The right decision depends on unit FOB value, process risk, labor content, and the number of days left before handover.

A useful commercial rule is cost versus time. If rework costs more than 35 to 45 percent of a clean remake and adds more than 5 to 7 days, remake is usually the lower-risk option. Typical FOB pricing helps frame that decision. A 1.25 inch soft enamel iron pin at 1.2 to 1.5 mm thickness commonly runs USD 0.24 to 0.48 FOB at 5,000 pieces and USD 0.18 to 0.34 at 10,000 pieces, depending on plating, backing, and carding. A 45 to 50 mm zinc alloy keychain with one body, one short chain, and one split ring often runs USD 0.62 to 1.10 FOB at 3,000 pieces and USD 0.52 to 0.95 at 5,000 pieces. A 50 mm die-struck coin, 3.0 mm thick, brass or zinc alloy, antique finish with standard edge detail, often runs USD 1.10 to 2.30 FOB at 1,000 to 3,000 pieces.

Lead times matter just as much as FOB. For common factories, a simple pin remake lot of 300 to 500 pieces usually takes 5 to 8 calendar days after final approval, including stamping, plating, coloring, drying, and pack-out. Keychain remake lots of 300 to 800 pieces usually take 6 to 10 days because hardware assembly adds a step. Coins often take 7 to 12 days depending on thickness, edge detail, and antique finish. If the defect is isolated to one plating batch or one SKU, a partial remake running in parallel with 100 percent sorting often protects the delivery window better than either full rejection or blind concession.

Concessions should be narrow and written like exceptions, not revised standards. If the client accepts backside haze on 600 coins to preserve launch date, the concession should state the exact defect, quantity cap, carton marks, SKU, photo reference, and confirmation that the exception does not apply to future runs. “Minor cosmetic issues accepted” is not a concession; it is an invitation to repeat the same failure.

4. Respec the process controls, not only the finished goods

A second failed inspection usually happens because the buyer corrected the symptom but not the process that created it. If enamel underfill keeps appearing around text strokes below 0.25 to 0.30 mm, the problem may be design-for-manufacture rather than careless filling. Fine isolated cells may need wider metal dams, deeper recesses, fewer segmented color islands, or a redraw with raised lines. If front-face haze or scuffing repeats, the root cause is often polishing inconsistency, poor pre-plating cleaning, rack contact marks, inadequate drying, or face-to-face friction after plating and before bagging.

Ask the factory to document process-side changes in writing. For stamped iron or brass pins, request die review, line depth target, polishing sequence, plating rack orientation, and final handling method. For zinc alloy keychains, request confirmation of die-cast gate trimming method, edge grinding sequence, hardware assembly control, and whether the ring is hand-closed or jig-closed. For antique coins, specify antique depth as light, medium, or dark against a sealed production sample because over-antiquing is often misread as dirty plating. Where attachment strength failed, require hardware source, wire diameter, and closing method before rework starts.

Numeric process controls prevent vague promises. Decorative nickel, imitation gold, black nickel, and gunmetal plating used on promotional items is typically appearance-grade, often around 0.03 to 0.10 microns, not jewelry-grade wear plating. Buyers expecting better abrasion performance on high-touch items should price a thicker spec or add protective packaging. For text and logo clarity, many factories struggle to hold clean enamel fill when stroke width drops below 0.30 mm or recessed cell width drops below 0.35 mm. For transit protection, mirror-polish pieces should not be bulk-bagged loose if faces can rub; specify one-piece OPP bagging, tissue interleaves, or tray cavities. “Improve plating” will not solve bag-rub damage created after plating.

5. Reset commercial terms, MOQ tiers, and recovery timing

Quality failures become expensive quickly: 100 percent sorting labor, replacement components, third-party re-inspection, repacking, re-cartoning, delayed balance payment, possible airfreight, and internal account-management time. Responsibility must be reset before corrective work starts. Otherwise the factory assumes one cost allocation, the buyer assumes another, and the argument starts only after rush work is already underway.

The corrective action sheet should state who pays for sorting, who pays for replacement parts, whether re-inspection fees are factory-borne or split, which quantity must be remade, whether remake runs in parallel with rework, and what release trigger authorizes partial shipment. If the defect clearly breaches an approved measurable spec, buyers commonly charge the re-inspection cost back to the factory. If the original spec was vague or changed after sample approval, a cost split is common because it keeps recovery moving instead of turning into a liability debate.

MOQ logic also needs realism. A factory may resist remaking only 180 failed units because plating setup, color mixing loss, drying rack utilization, and assembly yield make the run inefficient. Common workable remake tiers are 300 to 500 pieces for simple pins, 300 to 800 pieces for keychains with multiple hardware parts, and 200 to 500 pieces for thicker coins, depending on finish and packaging. In practice, authorizing the nearest efficient tier is often smarter than accepting borderline goods. Excess remake quantity can become service stock for shortage claims, client replacements, or distributor buffer inventory.

Corrective option	Typical added time	Typical cost impact	Best use case
100% sort only	1-3 days	Low to medium	Cosmetic defects are obvious, countable, and function is unaffected
Sort plus component rework	3-7 days	Medium	Jump rings, clutches, carding errors, or replaceable hardware issues
Partial remake	5-12 days	Medium to high	Failure is localized to one SKU, one lot, or one process batch
Full remake	10-20 days	High	Systemic finish, structure, or legibility problems across the lot
Accept with written concession	0-1 day	Low direct cost, high brand risk	Hidden-area issues only, with explicit quantity cap and carton traceability

6. Re-inspect against the new spec, SKU by SKU

The second inspection should be treated as a new control event, not a rerun of the original PSI. Provide the revised tolerance sheet, defect photos, updated AQL, and exact quantities that were sorted, reworked, remade, or conceded. If the factory claims 100 percent sorting, the inspector should verify segregation labels, rejected bins, carton traceability, and reverse checks on both passed and failed stock. A display table of “good samples” is not evidence that warehouse sorting actually happened.

For mixed orders, demand SKU-specific reporting. Pins, keychains, and coins fail in different ways, and one blended summary can hide the weakest SKU. Pins should be checked for overall size within drawing tolerance, thickness within plus or minus 0.10 mm where specified, post position, clutch fit, enamel fill level, edge burrs, and front/back plating appearance under consistent lighting. Keychains should include hardware count, jump ring closure, split ring pull or manual stress test, attachment alignment, swivel function where used, and plating contact marks. Coins should include diameter, thickness, edge detail consistency, antique contrast, front/back scratch review by zone, and rim deformation.

Pack-out must be verified as seriously as the product itself when the goods go direct to retail or event distribution. Carton count errors, mixed backing cards, skewed barcodes, and crushed hang holes trigger real chargebacks. Re-inspection should confirm unit count per inner, inner count per carton, carton gross-weight range, carton dimensions if booking is tight, and barcode orientation against the approved pack-out photo. If a partial release is planned, the accepted carton-number range should be listed on the release note so the forwarder and warehouse do not pull blocked stock by mistake.

7. Lock the lessons into the next PO and artwork approval

Once the shipment crisis is closed, the real value is preventing the next one. Update the master specification sheet, not just the complaint email thread. Add an annotated defect-photo appendix, appearance zones, hardware thresholds, finish description, packaging method, carton standard, and AQL by defect class. The next PO should inherit those controls automatically so the supplier is not restarting from a vague artwork approval.

For repeat custom metal products, maintain one live control file per SKU family. It should record approved artwork revision, material, thickness, finished size, plating finish, attachment type, carding method, pack-out standard, critical dimensions, tolerance limits, approved concession history, and non-repeat concessions. Useful examples include pin thickness 1.5 mm plus or minus 0.10 mm; coin diameter 50.0 mm plus or minus 0.20 mm; keychain overall length plus or minus 1.0 mm; jump ring wire diameter 1.8 mm minimum; text stroke width not below 0.30 mm unless specifically approved by process review; front-face zero-defect zone defined on an annotated image.

The best time to prevent a failed PSI is before mass production starts. Ask the supplier to flag likely production risks at artwork approval: fine text, isolated enamel cells, mirror-polish plating, mixed hardware, card insertion tolerance, and whether the finish will be bulk packed or individually bagged. A clean pre-production sample can hide scaling risk because real variation appears during plating handling, assembly, and pack-out at volume. The practical move is simple: turn the current failure into a numbered respec sheet before the next deposit is paid.

Have a project? Send your artwork and target quantity and we’ll reply with a detailed quotation within 12 working hours.