Knife defect classification for AQL: major, minor, and critical defects

Start with the defect decision rule

Write the decision rule before the defect list: critical first, then major, then minor. In our last 10 OEM knife POs, 6 named blade steel such as 3Cr13 or 5Cr15, carton-mark artwork, and the AQL level, but missed the classification rule. Bad start. The inspector walks in with a generic clipboard checklist, our line leader argues from the grinding line beside the 240# belt grinder, and the buyer receives 12 photos that still do not settle the case.

For knives, the rule follows risk. A critical defect knife issue is a safety or compliance failure, not a scratch debate. If the knife can cut the user during normal handling, or put the importer against LFGB, FDA, or REACH requirements, call it critical. We have failed lots for a folding knife lock that collapsed under firm thumb pressure on the bench, a blade tip punching through a 0.35 mm blister after carton-drop testing, metal swarf trapped in a PP handle, and food-contact test reports that did not match the actual resin lot. Most buyers set critical defects at AQL 0. One confirmed critical defect can fail the lot.

A major defect makes the knife unfit for normal sale or normal use. It does not need to injure the user on day one. It will still bring returns. At final QC last week, the grinding line sent up a chef knife with a 1.1 mm belly wave; it cut paper in one spot and skated in the next. We classify poor edge grinding, blade warp over the signed limit, a handle gap that catches a 0.20 mm feeler gauge, wrong EAN-13 barcode, or HRC outside the agreed band as major. Seven of nine European and North American importers we ship use AQL 2.5 for major defects. Downgrading these to minor just to pass shipment is the wrong move; the math doesn't work.

A minor defect is visible, but it does not affect safety, function, legal compliance, or the main brand face on shelf. Small stuff. A tiny polishing mark outside the main visual zone, a light box scuff on the back panel, or a logo position shift within the signed tolerance belongs here. QC pulled one sample this month with the logo 0.4 mm left of target; the buyer flagged it, but the approved drawing allowed ±0.8 mm. That stays minor. A common level is AQL 4.0. Write the rule before production, because if two inspectors classify the same defect differently, the defect list is not finished.

Spec line: blade material and hardness

Steel grade and heat treatment are spec lines we treat as inspection points, not decoration. They decide whether a blade still slices tomato skin after 30 cuts, whether salt spray leaves orange spots, and whether the color box claim survives a buyer audit. Write the check method beside the material on the PO: mill cert review, PMI spot check, or lab test for 5Cr15MoV, 7Cr17, 420J2, D2, AUS-10, 14C28N, or Damascus cladding. We see mix-ups on the grinding line. Last month the PO said 7Cr17, but the carton mark still showed 5Cr15MoV because the designer copied artwork from a 2023 order and nobody caught the line before carton printing.

Wrong steel grade is a major defect at minimum. It becomes critical when it creates a compliance or safety issue. A food-contact kitchen knife with an unapproved coating can sit at customs for 12 days vs 2 days for a clean file; a tactical knife declared with one composition but shipped with another leaves the importer facing regulators with weak paperwork. Visual inspection will not prove steel chemistry. QC can check stamping, finish, and rust under the bench light, but chemistry needs incoming mill certificates, PMI testing on selected lots, or lab testing on pre-shipment samples. We have seen this go sideways: the buyer flagged “D2” on the retail card, while the supplier certificate only covered 3Cr13.

Hardness should be written as a band, not a single target. “58 HRC” is the wrong question to ask. A chef knife spec should read something like 57–59 HRC measured at 3 points per sample, using a Rockwell tester after the blade cools from heat treatment and before final packing. For outdoor knives, 58–60 HRC can fit D2 or 14C28N, but the edge thickness behind the bevel matters: 0.35 mm cuts well, while 0.60 mm takes more abuse. Go below the lower limit and edge retention drops. Go above the upper limit and the grinding line starts hearing about chipped edges after the buyer’s drop test.

At TANGFORGE in Yangjiang, China, we do not approve a mass lot only from one golden sample if hardness supports your brand claim. Tie that sample to measured HRC, steel paperwork, and a repeatable check plan under AQL 2.5. For a 3,000 pcs kitchen knife order, QC pulled the sample, tested 3 points near heel, middle, and tip, then matched the result against the PO before we released packing. The math does not work if a 3,000 pcs lot ships on one untouched showroom sample.

Spec line: edge, tip, and cutting function

Sharpness is where loose inspection wording burns time. “Sharp enough” is not a factory spec. Put one test method on the buyer checklist and the inspection desk sheet. For normal shipment inspection, we run an 80 gsm paper slice test plus a 10X visual edge check under the bench lamp; CATRA stays for development samples or premium programs because it adds roughly 2 days and the fee does not fit every PO. Simple wins.

A dull edge on a premium chef knife is a major defect because the knife fails its main job. A small burr after the grinding line is major when QC pulled 4 pieces from a 32-piece AQL draw and found the same burr over 10 mm. It can stay minor only when it is one isolated piece and final honing is already written into the order. Edge chips are usually major. A broken tip can be major or critical depending on exposure and packaging. If the tip is broken inside a blister or sheath and can puncture packaging or injure warehouse staff, classify it as critical; we have seen this go sideways when the buyer flagged blood-risk handling during warehouse receiving.

For pocket, hunting, and tactical knives, point geometry and blade centering need hard limits written in mm. A point ground too thin, say under 0.25 mm behind the tip on a 58 HRC blade, can fail drop or torsion use. The math does not work if the tip looks sharp but bends in the first return test. A blade that contacts the liner during opening can damage the edge and create warranty claims; QC should check it with a feeler gauge, not just eyeball it. For folding knives, edge exposure when closed is serious. If a user can touch the sharpened edge through the handle gap in normal closed position, that is a critical defect knife case, not a minor workmanship issue.

Use measurable wording: no continuous burr over 10 mm, no edge chip over 0.3 mm, paper slice must cut 80 gsm paper from heel to tip without tearing, blade centering deviation not touching liner. These limits give the inspector something to apply instead of arguing at 5 p.m. before loading. The wrong question is “does it feel sharp”; write the mm limit, the paper grade, and the pass point on the QC sheet before we ship.

Spec line: handle assembly and lock safety

Handle and assembly defects get downgraded too easily because they look like small workmanship marks. Bad call. On the bench, these are the defects that turn into safety claims and returns. A loose handle scale on an 8 inch chef knife traps water after the buyer’s soak test; last month QC pulled 6 samples with dark residue at the rivet line after 20 minutes in warm water. A 0.25 mm gap at the bolster can miss food-contact hygiene expectations. A folding knife lock that slips can close on fingers. Same area on the knife, different defect class.

For fixed blade and kitchen knives, classify looseness by movement and hygiene risk. Any handle movement under normal hand force should be major; we check by hand first, then confirm suspect pieces with light pressure at the scale edge. A visible gap over 0.2–0.3 mm at the tang, bolster, or rivet area is major for food-contact knives because meat juice and detergent residue can sit there. Small cosmetic transitions between wood, G10, Micarta, ABS, or stainless parts stay minor only when the joint is smooth, sealed, and away from the palm-swell pressure zone. Use a 0.2 mm feeler gauge. Eye judgment starts arguments.

For folding knives, lock performance needs its own inspection line. Check liner locks and frame locks for face engagement, then check button locks, back locks, and axis-style mechanisms for accidental release under normal thumb pressure. Do not mix this with opening feel. If the lock fails during a reasonable spine pressure test defined by your QC plan, treat it as critical. If the blade has excessive side play but does not close, classify it as major; our grinding line has seen 0.6 mm side shake pass a quick visual check and fail as soon as the inspector held the tip. Slightly stiff action can pass as minor on a low-price utility knife. For a premium EDC order, the math does not work. The buyer will flag it.

Be careful with over-tightening as a factory fix. It hides blade play for 30 minutes. Then the customer gets a knife that opens like the pivot has sand in it. We see this after final assembly when a worker uses a T8 driver to chase the complaint instead of replacing the washer or checking the pivot stack height. A good defect list separates lock failure, blade play, pivot smoothness, screw stripping, handle cracks, rivet protrusion, and grip comfort; each item needs its own class and inspection method. One broad line called “assembly problem” is too loose for knives. We’ve seen this go sideways on AQL 2.5 lots.

Spec line: surface finish and branding

Surface finish is where buyers and factories argue most. Same mark, different result. One buyer passed a 4 mm hairline scratch on a budget camping knife, then rejected a 2 mm mark on a mirror-polished gift set because QC pulled the sample under the light box and the mark sat 6 mm from the logo. Set the rule before mass production. Your defect list should state viewing distance, light level, visual zone, and defect size in mm.

A shop-floor rule we run is to check visible surfaces at 30–40 cm distance under 600–800 lux without magnification. Main visual zones should include the blade face and handle top, with separate callouts for the retail-facing logo area and box front. The inspector should hold the knife at normal use angle, not hunt marks with a 10X loupe. Secondary zones can cover inner handle surfaces, spine underside, sheath back, and carton-facing parts. If the zone is not defined, the grinding line gets blamed for someone’s personal taste. Wrong question.

Branding needs its own spec line. Wrong logo, missing logo, wrong model number, wrong laser position, or private-label artwork that does not match the approved file should be major. If the barcode, FNSKU, carton mark, or country-of-origin label is wrong, classify it as major because Amazon receiving, retail DCs, or customs can stop the shipment at the dock. A 0.5 mm logo offset may be minor. A 3 mm offset on a 76 mm blade looks wrong fast, and the buyer will flag it if it breaks the approved drawing. We’ve seen this go sideways over one PO typo in the model number.

For cosmetic defects, don’t write only “small” or “obvious.” The math doesn’t work. Write limits: scratch length over 5 mm on the main blade face is major; pinhole coating defect over 0.5 mm is major; color difference outside the approved sealed sample is major; polishing line under 3 mm on a secondary zone is minor. If a knife has black oxide, stonewashed finish, PVD, or titanium coating, add adhesion checks and corrosion requirements, such as 3M tape pull on the coated logo area or salt-spray hours if the buyer requires it. Finish feels subjective only when the spec is lazy.

Spec line: packaging, labels, and compliance

Packaging defects often cost more than blade defects because QC catches them after the knives are already packed in master cartons. Reworking 5,000 boxed knives means cutting tape with a carton knife, opening cartons, pulling sleeves, changing labels, scanning EAN/UPC codes, then sealing again with 48 mm tape. Slow work. We’ve seen this go sideways when the packing table had 3 label rolls for the same SKU and the operator mixed old artwork with the approved version. In China export production, that rework adds 2–5 working days even when the blade, handle, and edge already passed inspection.

For knives, packaging is shelf display and safety control. Tip protectors, sheaths, edge guards, EVA inserts, clamshells, magnetic boxes, and warning labels all need to stop exposed points during trucking, container loading, and retail handling. A missing tip protector on a kitchen knife is normally major. A blade puncturing the retail box or polybag should be critical because it can cut warehouse staff or consumers; QC pulled one sample last year where the 1.8 mm tip came through the corner of a color box. If your market requires warnings, age labels, Proposition 65 language, recycling symbols, food-contact statements, or importer address, missing required text is major and sometimes critical from a compliance angle. Don’t “fix later” here.

Carton information should match the packing list and purchase order: SKU, quantity, gross weight, net weight, carton size, country of origin, batch number, and shipping marks. The buyer flagged it once because the PO said “matte black handle” and the carton mark said “black matt handle.” Small typo, big hold. For DDP or retail DC delivery, barcode readability is not optional. We run scan testing on at least 5–13 units per SKU during final inspection, with extra scans if the packing line uses 2 label versions or changes rolls mid-shift. A $35 handheld scanner at the QC desk catches problems before the DC rejects the pallet.

Common classification: crushed master carton affecting product protection is major; light carton scuff is minor; wrong FNSKU is major; mixed SKU inside one inner carton is major; missing silica gel when specified for carbon steel or Damascus knives is major if corrosion risk exists. The math doesn’t work if paperwork waits until shipment day. Compliance documents should be tied to inspection: LFGB/FDA food-contact reports, REACH/SVHC declarations, BSCI or ISO 9001 documents, and test reports must match the actual product, not just the supplier name. We check model code, blade material, coating, handle material, and report date against the sealed sample before release, and QC writes any mismatch on the AQL sheet instead of leaving it for the shipping clerk.

Build the list inspectors can apply

A usable defect list sits on the inspection table, not three pages deep in an email chain. For knife programs we ship, 40–70 defect lines usually cover the real risks; once the sheet passes 100 lines, QC spends more time reading than checking blades, unless a retail compliance team demands that format. We run ours in a spreadsheet: spec line, defect wording, class, measurement method, sample reference, photo example, and rework decision. Simple wins. Last month QC pulled a 210 mm chef knife sample where “handle gap” had no limit. The buyer flagged 0.25 mm, production accepted 0.50 mm, and that gap debate burned 2 hours beside the packing belt.

Set AQL by commercial risk, not panic. A common setup for export knives is Critical AQL 0, Major AQL 2.5, Minor AQL 4.0 under ANSI/ASQ Z1.4 or ISO 2859-1 single sampling, general inspection level II. For first orders, new tooling, or a new steel batch, tighten major defects to AQL 1.5. For stable repeat orders, keep 2.5 and add line checks where defects start: blade thickness at grinding, hardness after heat treatment, and barcode/carton checks before final packing. We’ve seen this go sideways. The blade passes final inspection, but the grinding line already left a 0.3 mm over-grind near the heel, and nobody wrote whether that mark counts as major or minor.

At TANGFORGE, our monthly capacity is about 300,000–500,000 knives across kitchen knives, outdoor fixed blades, pocket knives, and Damascus items, with typical OEM MOQ from 600–1,200 pcs per SKU depending on material and packaging. Classification is not office paperwork for us. It protects the ship date. On a 4,800-piece order, one loose phrase like “scratch on bolster” can hold 96 cartons, take two pallet positions in the warehouse, and push the vessel booking from China by 7 days. The math does not work when the PO says “minor cosmetic acceptable,” but the incoming inspector rejects every 8 mm hairline mark under a 6000K LED lamp.

Before mass production, approve one golden sample, one limit sample for acceptable cosmetic variation, and a reject photo sheet with close-ups and mm limits. Send the same file to factory QC, the third-party inspection company, and your receiving warehouse; nobody should judge a handle color from an old WeChat screenshot. If a defect class changes after production starts, record the version number and effective date. We once had a buyer’s PO show “black POM handle,” while the signed sample card said “dark brown pakka,” and QC stopped the line until purchasing confirmed it. Inspectors do not need polished wording. They need table-side steps: check the part, measure with the caliper, classify it, count it, then pass or fail the lot.

Frequently asked questions