How to Classify Knife Defects for AQL Inspections

Start with risk, not opinion

The cleanest way to build knife defect classification is to start with consequence. A critical defect knife issue means the user can get cut, the product fails food-contact rules, or the knife becomes unsafe or illegal to sell. A cracked blade, a folding knife that fails lock-up, a handle that separates under 15 kg hand-pull pressure, or black polishing compound trapped at the blade-handle gap all belong in critical. QC pulled one sample last year with a 0.8 mm crack near the heel after the Rockwell bench check. No debate. Scrap it.

Major defects are different. They do not usually hurt the user right away, but they damage cutting performance or shelf acceptance. Blade warp over 1.5 mm, an uneven grind from the wet belt, a burr that tears A4 paper instead of cutting cleanly, loose handle scales that have not separated, or a wrong barcode on the color box can all become major. The buyer flagged this once on a 3,000 pcs shipment because the EAN code on the PO had one wrong digit. Minor defects are low-impact cosmetic issues: a 3 mm scratch on the back side of the handle, a laser mark shifted 1 mm off center, or a carton scuff away from the retail face.

The mistake 7 out of 10 new buyers make is treating every visible issue as major. This is the wrong question to ask. It pushes rejection rates up, and the supplier ends up arguing over photos instead of fixing the defect that matters. A practical list should state the defect, measurement, location, and viewing condition, such as 30 cm distance under normal white inspection light. If an inspector needs to guess whether a stain is serious, the list is too weak. We have seen this go sideways during AQL 2.5 inspection when one buyer called a 1 mm oil dot on the spine “major” but ignored a loose rivet. In Yangjiang, China, good factories do not want arguments either; we want a line item the grinding line can sort, rework, or scrap before packing.

Build a usable defect list

The fastest way to stop mixed inspection calls is to write each defect as a sentence with a hard limit. Don’t write “scratch” or “gap” by itself. Write “scratch longer than 10 mm on the show side of the blade under normal room light” or “handle gap wider than 0.3 mm at the bolster.” Then QC can check it with a steel ruler, a 0.3 mm feeler gauge, and the same 600–800 lux bench light we run at final inspection.

This is where 6 out of 10 new programs go wrong: the team classifies by feeling instead of evidence. Wrong question. A polished stainless chef knife with a 2 mm surface mark near the spine is not the same as a 2 mm mark on the cutting edge. The math does not work if both get the same call. QC pulled a sample last month where the buyer flagged a tiny anodizing scuff as major on a satin blade show face because the retail spec allowed no visible marks after blister packing. Inspection criteria knife decisions should point to the knife type and finish, then tie the call to the customer standard and the viewing distance written on the PO.

Set rules by knife type

One defect list fails at the packing table. A kitchen knife, a folding knife, and a Damascus display piece usually fail at different stations, so we run separate inspection blocks instead of pushing one generic checklist across all SKUs. Last month QC pulled 32 samples from a 1,200 pcs mixed PO: the chef knives had bevel width drift after the grinding line changed belts, while the folders showed liner rub near the pivot. Same AQL. Different pain.

Kitchen and chef knives

For kitchen knives, check the edge first: bevel symmetry within 0.3 mm, no burr on the cotton wipe test, a straight blade spine, handle gaps under 0.2 mm, and food-contact files that match the approved sample. A bad grind, uneven bevel, burrs, or tea-colored staining on 3Cr13 or 5Cr15 stainless steel belongs under major defects because cutting feel and corrosion claims land on the buyer’s desk fast. If the item is sold for food use, keep LFGB, FDA, and REACH paperwork tied to the physical sample; we have seen a shipment rejected because the PO said “satin finish” while the counter sample was mirror polish.

Pocket, outdoor, and tactical knives

For folding knives, arguing about tiny handle dots is the wrong question to ask before the lock is checked. Lock-up, blade centering within about 0.5 mm, smooth opening action, pivot screw torque, and covered tip position decide whether we ship or rework. If the blade touches the liner, the lock slips under spine pressure, or deployment feel changes across samples, we move it from major toward critical fast. On sheathed outdoor knives, sheath retention needs a pull check with the belt clip fitted; one buyer flagged a batch where 6 out of 80 samples slid loose when the carton was tilted.

Damascus and gift sets

Damascus knives and premium gift sets need tighter visual rules because the buyer pays for the look before anyone cuts an onion. Pattern breaks, weak etching near the heel, glue squeeze-out inside the box, and a crooked sleeve logo should not be treated like small rub marks on a bulk chef knife. The math does not work. On a 500 pcs gift-set order, even 15 sets with uneven acid etch can trigger retail complaints, especially when the finish was approved from a hand-picked golden sample under the light box.

Map defects to AQL levels

AQL is not the defect class. It is the accept/reject limit you assign after the risk is named. Critical means Ac 0 on our floor. One loose blade tip, one exposed rivet burr, or one food-contact contamination mark stops the lot before packing continues. Major defects usually sit at AQL 1.0 or AQL 1.5. Minor defects go to AQL 2.5 or AQL 4.0, depending on buyer channel and product grade. Simple rule. At the packing table, we check with 0.02 mm calipers, edge guards, and the signed defect list, because a soft AQL sheet turns into an argument after QC pulls the sample.

Do not hard-code the sample count without ISO 2859-1 and the lot size. This is the wrong question to ask. A 5,000-piece order can use a different code letter from a 50,000-piece order, even with the same AQL. We ship about 120,000 units per month from Yangjiang, and the sample size decides whether the grinding line waits 2 hours or the full shipment sits until the next day. On a first shipment, we run the major limit tighter and ask for photo evidence on every rejected carton, including label mistakes like a PO typo on blade size. For repeat orders, loosen the program only after 3 clean lots in a row.

Inspect what actually fails

Knife inspection should focus on defects that actually show up on the grinding line and packing table, not on a 40-point wish list written in an office. We run calipers, a Rockwell hardness tester, a 5x loupe, a flat reference plate, and a quick paper cut test to catch edge skips. For chef knives, check blade straightness against the agreed mm tolerance, grind symmetry, tip alignment, handle gaps, plus the bolster area where polishing wax hides in the corner. For folding knives, check lock engagement, blade centering, pivot tension, and accidental opening risk. Simple rule. “Can the knife still sell?” is the wrong question to ask. The real question is whether the buyer will reject the lot after QC pulled 125 pcs under AQL 2.5.

• Incoming steel: verify grade and thickness, then match the heat treatment record to the batch card before blanking
• In-process: check bevel angle, tip formation, and handle fit before final assembly, because fixing a 0.6 mm handle step after riveting eats time
• Pre-shipment: inspect edge condition, rust spots, label accuracy, and carton count; one wrong SKU sticker can hold 3 pallets at the warehouse

Sampling needs cartons from the top, middle, and bottom of the pallet. One bad packing layer is enough. We have seen this go sideways when 18 cartons in the center layer carried the old PO revision, while the outside cartons looked clean. If the order ships to Amazon, verify carton marks, FNSKU, and unit barcode readability with the same scanner the packing team uses, not just by eye. For food-contact knives, a rust dot near the heel that measures only 1 mm in the factory can still become a return claim after sea freight and warehouse storage. The inspector should photograph each fail with a ruler, fixed light angle, and lot number visible in the frame. Then the report is traceable, not a debate over whose screen looks brighter.

Write the list the factory can run

A defect list works only when the line leader can use it at the packing table without calling 6 people over. Each row needs a plain defect name and severity, the limit in mm, HRC, or viewing distance, the QC station, one photo from the approved sample, and the action: rework, scrap, or hold lot. Be blunt. “Handle gap over 0.30 mm at bolster, Major, check at final QC with feeler gauge, rework allowed before packing” is clear. “Poor fit” is useless. QC pulled the sample. We’ve seen this go sideways when a buyer flagged black spots on pakkawood after packing, but the list never said whether 1 spot under 1 mm was acceptable.

At our Yangjiang, China operation, we build around MOQ 500 pcs for roughly 40 custom programs a year, lead time 35 to 45 days after sample approval, and common hardness bands from HRC 56 to 61 depending on the blade design. Those numbers protect nobody if the inspection rules are loose. A contradictory spec burns time under FOB, DDP, or third-party warehouse handling in Zhejiang; last season one PO said “mirror polish,” while the approved sample was satin #400 from the grinding line. The buyer flagged it late. The math doesn’t work when QC pulls 80 pcs at AQL and the team argues about finish instead of counting blade scratches, handle gaps, or logo drift.

Keep revision control strict. If the steel changes, reopen the defect list. If the handle resin changes or the buyer asks for a new finish, reissue the golden sample and mark the date on the carton file. A clean list is not office decoration. It keeps production and the importer judging the same knife before we ship the container out of China, down to small checks like 0.20 mm tip bend, logo position, and whether the edge guard scratches the blade during transit. We run this check before mass packing, not after 38 cartons are taped shut.

Frequently asked questions