A knife shipment can pass in the sample room and still turn into a claim 12 days after landing: chipped 0.3 mm tips, loose POM handles, crushed 5-ply cartons, EAN-13 barcode mismatch, or blade hardness drifting outside the PO spec. Small misses cost money. Most claims we see are not one dramatic failure; QC pulled the sample, found one unchecked line on the inspection sheet, and the buyer flagged the same issue across 48 cartons.
At TANGFORGE in Yangjiang, Zhejiang, China, we build OEM and ODM knives for importers and brands that need repeatable inspection, not lucky batches. We run about 300,000 units per month across kitchen, chef, outdoor, pocket, hunting, tactical, and Damascus knives, with the grinding line checking edge angle by gauge before final wipe-down. This guide reads like a working spec sheet: what to check, why it matters to your shipment, and where final inspection should stop the goods before someone starts explaining defects by email.
The QC Spec Starts Before Cutting Steel
A knife QC guide should start with the purchase spec, not the inspector’s checklist. If the PO says only “8 inch chef knife, pakkawood handle, retail box,” the factory merchandiser, the grinding line, and the third-party inspector will read three different products from one line. That is how claims start. Write acceptance limits for each point that touches safety, cutting performance, compliance, or shelf look, such as “blade length 203 mm ±2 mm” and “carton drop test 76 cm, 1 corner and 3 edges.”
For a kitchen knife, the spec should name the blade steel grade and target HRC, then lock the working dimensions: blade length tolerance, spine thickness at heel in mm, grind type, and edge angle per side. Handle details need the same treatment. Call out handle material, rivet or fastener type, logo position with a mm reference from the bolster or butt, surface finish grade, retail box structure, barcode format, and carton loading. For a pocket or tactical knife, add lock type, opening force in N, detent feel benchmark, clip torque, liner thickness, closed length, and the legal limit for the sales market. QC pulled one folding sample last month because the PO said “strong clip” but gave no torque number; the buyer flagged it after the clip bent in a jeans-pocket test.
The buyer impact is blunt: if a line is not written, rejection gets messy. This is the wrong question to ask: “Will the inspector know what good looks like?” Ask whether the inspector has a signed number or approved sample to compare against. A quality inspection company can check blade runout with calipers, read hardness on the HRC tester, and scan EAN-13 codes, but it cannot invent your brand standard. We run a golden sample for hand-feel and visual balance, then a signed technical sheet for numbers such as 60-62 HRC, 15° edge angle, or AQL 2.5.
At TANGFORGE in Yangjiang, China, we keep approved samples, pre-production samples, and final shipment samples on separate QA room shelves with red, yellow, and green labels. Simple system. It stops the common mistake of checking final goods against an early salesman sample instead of the locked production sample. For OEM orders, our standard MOQ often starts at 500 pieces per SKU for simpler kitchen knives and 1,000 pieces for more complex folding knives, depending on tooling and packaging. Small batches still need the same discipline because a 3% defect rate on 500 pieces is 15 customer complaints; we have seen this go sideways when a PO typo changed “matte handle” to “metal handle” and nobody caught it before pack-out.
Incoming Inspection Lines That Prevent Rework
Incoming inspection is not paperwork theatre. It is the last low-cost stop before labor, grinding, heat treatment, assembly, and packing turn one bad lot into 3,000 bad knives. We run the check at the receiving table with a caliper, magnet, barcode scanner, and the supplier PO beside the carton. For knives, the line covers steel strip or blanks, handle scales, screws, rivets, abrasives, retail packaging, inserts, master cartons, labels, plus outsourced parts such as sheaths or nylon pouches.
The first line is material identity. For steel, the factory should match the supplier mill certificate to the PO and run PMI or chemical composition checks on risky lots. If your spec is 1.4116, 420J2, AUS-8, D2, 14C28N, VG10, or 10Cr15CoMoV, do not let anyone write “stainless steel” on the inspection sheet. Wrong answer. QC pulled a sample last year marked 5Cr15 when the buyer ordered 14C28N; it looked clean after polishing, then failed the 24-hour salt spray check. The wrong steel can pass pack-out and still come back as corrosion claims or weak edge retention after retail sale.
The second line is dimension. Blade blanks need checks for length, width, thickness, tang shape, hole position, and warp, with readings written in mm instead of “OK.” A 0.3 mm hole shift can make a handle scale sit proud. A 0.5 mm blade thickness drift can change weight, balance, grind time, and final edge feel. The grinding line feels this fast. Handle material also needs moisture, crack, color, odor, and thickness checks; we use a moisture meter on wood and pakkawood before assembly. Wood and pakkawood should be stable before assembly because wet stock can shrink after shipment and leave gaps at the bolster or rivets.
Packaging gets ignored until final inspection, and that is the wrong question to ask at 18:30 with a container booked for the next morning. Retail boxes, magnetic gift boxes, sleeves, instruction cards, FNSKU labels, and master cartons should be checked when they arrive, not after 2,400 knives are packed. We scan barcodes with a handheld Zebra scanner and drop-check the carton sample before it reaches the packing tables. A knife can be perfect, but if the barcode scans to the wrong SKU or the carton collapses in transit, the importer still pays. For North America and Europe, we also check REACH, LFGB, FDA food-contact declarations, and customer-specific labeling before mass pack-out.
Blade Geometry, Hardness, and Edge Data
The blade is where end users spot trouble first. Asking “is it sharp?” is the wrong question to ask. We can make a weak edge shave paper after one hard pass on the buffing wheel, then QC pulls the sample two days later and the bite is already gone. The better spec lines are blade geometry, heat treatment, edge consistency, and rust behavior after washing.
Write hardness as a band, not a single hero number. For German-style kitchen knives using X50CrMoV15 or 1.4116, 56–58 HRC is common. For Japanese-style VG10 or 10Cr15CoMoV chef knives, 59–61 HRC is common. For D2 outdoor knives, 58–60 HRC is often requested. Harder does not mean better. We have seen 62 HRC kitchen knives come back with chipped edges after 18 days in normal home use, while the same pattern at 58 HRC ran 12 months with fewer claims. On the Rockwell tester, we run 3–5 points per lot and mark the heel area before polishing so nobody hides a bad reading under a mirror finish.
Edge angle needs a tolerance your grinding line can actually hold. A common kitchen knife spec may be 15° per side with ±2° tolerance. Outdoor knives may use 20° per side or a convex edge depending on use. If left and right bevels drift by 4° at the tip, the knife can still pass a quick paper cut test, then feel dead after one week on a plastic board. For higher-volume programs, CATRA testing gives better edge retention data, but the math does not work for every reorder. We run CATRA for validation on new SKUs, then use angle gauges, in-line edge checks, and final sharpness sampling for routine shipments; last month the buyer flagged a PO typo that said 12° instead of 15°, and QC stopped 600 pcs before final pack-out.
| Spec line | Typical check | Buyer impact |
|---|---|---|
| Hardness | 3–5 points per lot, target ±2 HRC | Cuts chipping and soft-edge claims |
| Blade thickness | Caliper at heel, middle, tip | Controls weight, balance, and grind consistency |
| Edge angle | Angle gauge or optical check | Keeps cutting feel and resharpening consistent |
| Tip alignment | Visual jig and flat reference | Prevents bent-tip returns and safety complaints |
| Corrosion | Salt spray or customer soak test | Backs compliance claims and cuts rust complaints |
On our Yangjiang production lines, HRC is a process result, not a shiny catalog claim. If a batch misses the band, polishing and packaging stop. We check the heat-treatment chart, furnace lot number, quench timing, and retest data before that pallet moves to the packing table.
Handle, Rivet, Lock, and Assembly Fit
Assembly defects trigger returns because the customer feels them in the first 5 seconds. A 0.20 mm gap between handle and tang, a proud rivet, a loose scale, or gritty folding action makes the knife feel cheap even when the blade steel is right. Your QC sheet needs two lines: cosmetic tolerance and functional tolerance. They are not the same. QC pulled 32 samples last month where the satin finish passed, but the rear rivet sat 0.12 mm proud and caught a cotton glove.
For fixed-blade kitchen and outdoor knives, check handle alignment, scale flushness, epoxy squeeze-out, rivet height, bolster transition, spine rounding, and butt finish with a feeler gauge and fingertip pass. A workable handle gap standard might be no visible gap over 0.15 mm on premium chef knives and no sharp edge on any handle contact surface. For lower-price utility knives, wider cosmetic tolerance can pass, but loose parts cannot. The math doesn't work if a buyer saves USD 0.08 on handle finishing and then gets 3% returns. We run a simple thermal cycle, 2 hours warm and 2 hours cool, then press the scale by hand; the handle should not click, shift, or open a glue line.
For folding knives, the inspection sheet must include blade centering, lock engagement, lock release force, opening and closing smoothness, pivot screw torque, pocket clip tension, and blade play. A liner lock that engages at 10% feels unsafe; one that travels to 90% can wear out fast. Several US buyers we ship to set the acceptable engagement window around 30–70%, but the exact line depends on the lock geometry. On the grinding line, the buyer flagged 18 pieces because the blade sat 0.8 mm off center after the pivot was adjusted for smooth action.
Fastener torque gets missed too often. If screws are too loose, they back out during transit or first use. If they are too tight, they strip threads or kill the action. For repeat programs, define torque in N·cm for pivot, clip, and scale screws, then put the value on the QC sheet instead of leaving it to the assembler's feel. Threadlocker color and amount also need a limit; we have seen blue adhesive squeeze into washers and make 240 knives feel gritty before pack-out.
Final inspection needs a tactile pass by trained QC staff, not only a visual pass. Hands find proud edges, wobble, grit, and bad balance faster than a photo report. This matters on ODM designs where the buyer has not handled 3,000 production pieces yet. At final pack-out, we ask QC to open and close every folding sample pulled under AQL 2.5, then wipe the handle with a white cloth so burrs, oil bleed, and black polishing compound show up before we ship.
Appearance Defects Need Named Limits
“No scratches” sounds strict. It is the wrong question to ask. A mirror-polished chef knife and a bead-blasted tactical knife cannot share one cosmetic line on the QC sheet. The spec needs the defect name, max size in mm, allowed location, and viewing condition, or QC pulled the sample and everyone starts fighting over WeChat photos after 24 cartons are taped shut.
Knife appearance defects we see at pack-out include blade scratches over 3 mm, cross grind marks from the 400# belt, broken satin lines, pitting, black spots, rust, burrs, logo shift, laser burn, handle cracks, color mismatch, glue residue, dirty sheaths, crushed boxes, and wrong inserts. Location changes the call. A 0.5 mm dot near the tang can pass as minor on a budget outdoor knife. Put that same dot on the front face of a premium Damascus gift knife and the buyer will flag it as major because the shelf photo looks cheaper.
Set the viewing rule in writing. About 7 out of 10 import QC specs we receive use 30–40 cm viewing distance under normal white light for cosmetic inspection. Premium retail knives sometimes get a light box check at 6500K, with the blade held flat for 5 seconds per side. Be careful with impossible standards. If you demand zero visible micro-scratches on a hand-polished blade at 10 cm under strong light, the grinding line will lose yield and the price will move. That math works for a $90 FOB premium set. It does not work for a $6 FOB promotional knife.
Logo and marking defects need their own line. Laser engraving should be checked for position, size, darkness, spelling, and orientation; we once had QC stop 1,200 pcs because the PO wrote “stainess” instead of “stainless” on the artwork file. Etched logos should survive a simple wipe test with a damp cloth. If you sell through Amazon or retail chains, wrong barcode, wrong FNSKU, wrong country-of-origin label, or mixed SKU cartons are major defects, not office mistakes. A warehouse rejection costs more than a blade scratch.
Good factories in China can meet tight cosmetic standards when the price, process, and inspection time support them. We run tighter checks when the buyer pays for extra polishing, clean PE bag handling, and AQL time at final inspection. Be honest on the spec. Write the standard your market needs, not the one that sounds good in a sourcing meeting.
AQL Sampling and Final Inspection Rules
Final inspection is the shipment gate. It checks production against the approved spec; asking QC to fix unclear specs at this stage is the wrong question to ask. About 8 out of 10 importers we work with call out ANSI/ASQ Z1.4 or ISO 2859-1 sampling at General Inspection Level II. A common setting is still AQL 0 for critical defects, AQL 2.5 for major defects, and AQL 4.0 for minor defects. For premium knives or first orders, we run tighter sampling or 100% checks on the risky points: lock release force on the jig, barcode scan accuracy with a handheld scanner, and blade exposure after pack-out.
Critical defects are safety or legal failures: exposed sharp edge cutting through the blister, broken liner lock, blade tip sticking out of the sheath by 1 mm, wrong steel declaration on the label, missing warning text, or oil and dust contamination found when QC pulled the sample. Major defects hit saleability or function: loose handle with visible gap over 0.3 mm, dull edge that fails the paper-cut test, wrong logo position, blade warp seen on the flat plate, rust spots, wrong carton mark, failed drop test, or mixed SKUs in one master carton. Minor defects are small cosmetic issues inside the written limit, such as one tiny non-front-facing mark that retail buyers will not reject.
For a 5,000-piece lot under General Level II, sample size code L often means 200 pieces inspected. At AQL 2.5, the accept number is typically 10 and reject number is 11 for major defects. Do not rely on memory. We have seen this go sideways when an inspector copied “code K” from an old PO typo and checked 125 pieces instead of 200. The report should state the sampling plan, then show defect photos, quantities, carton numbers, caliper measurements in mm, function test results, packaging checks, plus a clear pass, fail, or pending status.
Final inspection should happen when at least 80% of goods are packed and 100% are produced. Inspect at 50% packed and you will miss pack-out issues like wrong inner box direction, weak tape, or barcode labels applied to the wrong color SKU. Inspect after the container is booked with no buffer and the math does not work; a 2-day rework becomes 5 days once the grinding line and packing team are already on the next order. We advise importers to reserve 2–3 working days between final inspection and loading, and for first production in China or new packaging, we prefer 12 days of shipping buffer rather than forcing a shipment date that needed 18 days from the start.
Pack-Out Checks That Stop Claims
Pack-out is not just sliding a knife into a box. This is where a passed blade turns into a claim. We have seen 600 pcs of good chef knives fail sell-through because the sleeve corners crushed and the hang tags rubbed black in the master carton. A sharp knife needs a fitted edge guard, dry packing material, readable labels, tight inner boxes, and export cartons that can take forklift handling. If pack-out is skipped, the steel may pass 58 HRC and the edge may look clean, but the retail pack still lands unsellable.
Start with safety. Blade guards, tip protectors, sheaths, clamshells, or magnetic boxes must stop the edge from cutting through packaging during vibration. For chef knives, QC should press the tip area against the sleeve by hand because the first puncture usually starts within 8 mm of the point. For outdoor knives, we invert the sheath sample 10 times over the packing table; if the knife drops out and the spec did not call for quick draw, the math doesn't work for export cartons.
Check moisture and corrosion control next. Stainless is not rust-proof, and buyers still push back on this after the first spot appears near the logo etch. If goods are polished, washed, packed warm, and sealed too fast, trapped moisture can create rust spots during a 30–45 day ocean shipment. We run dry cartons, clean polybags, rust-inhibiting paper where the spec calls for it, and a 24-hour hold before loading when the grinding line has just washed blades. For high-carbon Damascus or carbon steel knives, light oil and instruction cards belong in the packing spec, not in an email after shipment.
Carton checks matter for claims and chargebacks. Confirm inner quantity and master carton quantity against the PO, then check gross weight, net weight, carton size, shipping marks, PO number, SKU, barcode, FNSKU if required, and country of origin. Last quarter QC pulled a sealed carton where the barcode was right but the PO had one digit typed wrong; that would have become a warehouse rejection. A simple carton drop test from 76 cm is common for export packaging validation, though exact height depends on carton weight and customer standard. For retail distribution, ISTA-style testing is often cheaper than arguing over crushed corners later.
At TANGFORGE in Zhejiang, China, final pack-out checks include scan verification, carton count reconciliation, and random box opening after sealing. We scan 100% of FNSKU cartons when the buyer requires Amazon routing, then open 2 sealed cartons per lot to check label face, inner count, and blade guard position. It is boring work. That is the point. Boring QC prevents exciting emails from angry customers.
Frequently asked questions
For most OEM knife shipments, use AQL 0 for critical defects, AQL 2.5 for major defects, and AQL 4.0 for minor defects under ANSI/ASQ Z1.4 or ISO 2859-1, General Inspection Level II. Critical defects include unsafe locks, exposed blade tips, wrong legal labeling, or contamination. Major defects include loose handles, rust, wrong logo, dull edges, or failed packaging. For first orders, premium knives, or folding knives with lock systems, tighten the plan or add 100% function checks on the highest-risk items. The important point is to define the AQL in the PO before production, not after defects appear in the final inspection report.
For routine production, many buyers check 3–5 pieces per lot or per heat-treatment batch, depending on batch size and steel type. If the lot is large or the steel is high-risk, increase sampling. The spec should show a target band, such as 56–58 HRC for many 1.4116 kitchen knives or 59–61 HRC for VG10-style chef knives. A single HRC reading is not enough if the batch includes multiple furnace loads. Ask the factory to keep heat-treatment records and identify which cartons or production dates belong to each batch. If any reading falls outside the band, pause shipment and retest before pack-out.
The most expensive claims usually come from functional and pack-out defects, not tiny cosmetic marks. Common problems are chipped tips, loose handles, dull or uneven edges, rust spots, blade warp, folding lock failure, wrong logo, wrong barcode, mixed SKU cartons, crushed retail boxes, and knives cutting through packaging. Cosmetic defects still matter, especially for premium gift sets, but a warehouse rejection for wrong FNSKU or a safety complaint for a failed lock is worse. Your inspection checklist should classify defects as critical, major, or minor with photos and size limits. That gives the inspector authority to stop shipment before the product leaves China.
Both can make sense. The factory should always run incoming inspection because it controls daily production and can reject material before assembly. A third party can add value on first orders, high-value shipments, or when your own QA team cannot visit China. For example, the factory checks steel certificates, handle moisture, cartons, and labels on arrival; your third-party inspector later verifies the same risk points during final inspection. Do not use third-party inspection as a substitute for factory process control. If wrong handle material enters production on day 1, discovering it on day 25 means rework, delay, and avoidable cost.
Schedule final inspection when 100% of the order is produced and at least 80% is packed. This timing allows inspectors to check real pack-out conditions, carton marks, barcode scans, counts, retail box condition, and random sealed cartons. Leave 2–3 working days before container loading so failed items can be sorted, repacked, or reinspected without forcing a bad decision. For new SKUs, complex gift packaging, or folding knives with lock checks, add more buffer. If you book inspection on the morning of loading, you are not managing quality; you are gambling with the shipment schedule.
Send Us Your Knife QC Sheet
Share your drawings, target AQL, packaging spec, and market requirements. Our Yangjiang team will review the inspection risks before sampling or mass production.
Request a Quote
