What your factory actually does when you spec AQL 2.5

What does AQL 2.5 actually mean to the people running your inspection?

AQL 2.5 means if the number of major defects found in a random sample exceeds the acceptance limit set by ISO 2859-1 tables, your entire lot gets rejected. It does not mean 2.5% of your shipment is allowed to be defective. Most brands misunderstand this distinction, and that misunderstanding creates friction between brand and factory.

I run quality operations inside a vertically integrated factory in Fuzhou. When your tech pack lands with "AQL 2.5/4.0" written in the QC section, here is what actually happens on our floor.

How does a factory set up AQL inspection before production starts?

We set up the inspection framework before a single garment gets cut. The production manager reviews your spec sheet and identifies which defects fall into critical, major, and minor categories for this specific style. A loose thread is minor. A measurement 2cm outside tolerance is major. A needle left in a hem is critical.

We print a defect classification sheet in Mandarin and English and post it at every QC station. The inline checkers and final inspectors need to sort defects the same way. If your brand uses a custom defect classification, we need that document before production. Otherwise, we default to industry standards.

The inspection plan also specifies sample size. We use the ISO 2859-1 tables at General Inspection Level II for most orders. A lot of 5,000 units requires a sample of 200 pieces. The sample gets pulled randomly across cartons, not just from the first few boxes.

Where does inline QC actually happen during production?

AQL inspection at shipment is the last line of defense. The real work happens earlier.

We run inline checks at three stages:

• After cutting: Checkers verify panel counts, fabric direction, and match between pattern pieces and cut pieces. Fabric defects get caught here. Slubs, shading, holes. If a roll has excessive defects, we pull it before it becomes sewn inventory.

• During sewing: Roving inspectors walk the lines and pull random pieces off machines. They check stitch density, seam alignment, and construction accuracy against the approved sample. Problems get flagged immediately while the operator can still adjust.

• After finishing: Pressing, folding, labeling. Inspectors verify care labels match declared content, hangtags are correct, and packaging meets spec.

By the time a lot reaches final inspection, most defects have already been caught and reworked. Final AQL inspection confirms the system worked.

"The myth is that QC is done only at the end, a second before it leaves the factory floor. But by then, it's too late to address anything."

What happens during final AQL inspection?

Final inspection happens when at least 80% of the order is packed and cartooned. We do not inspect incomplete lots. If you push for inspection at 50% completion, the sample will not be representative.

The inspector pulls cartons using a random number sequence. From those cartons, they pull individual units. For a 200-piece sample from a 5,000-unit lot, we might pull from 40 cartons, taking 5 units from each.

Each unit gets checked against a standardized checklist:

• Measurements on 5-10 points depending on garment type
• Visual inspection for stains, holes, uncut threads, skipped stitches
• Construction verification against the sealed sample
• Label and hangtag accuracy
• Packaging condition

Defects get logged by type and severity. The inspector tallies critical, major, and minor defects separately. Then they compare against the acceptance numbers from the AQL table.

For AQL 2.5 on a 200-piece sample, the accept number for major defects is 10 and the reject number is 11. Find 10 or fewer major defects, the lot passes. Find 11 or more, the lot fails.

Why do factories sometimes disagree with brand QC teams on defect classification?

This is where partnerships break down.

A brand's third-party inspector arrives and classifies a slight color variation as a major defect. Our internal team classified it as minor because it falls within the approved lab dip tolerance. Now the lot fails their inspection but passed ours.

The fix is alignment before production. Send us your defect classification document. Tell us which issues are deal-breakers for your customer base. If your market tolerates minor thread tails but rejects any measurement variance, we need to know that.

I worked with a Melbourne activewear founder last year who spent three months arguing with their previous factory over what constituted "acceptable pilling" on a brushed jersey. The factory thought light pilling was inherent to the fabric. The founder thought any pilling was a defect. Neither was wrong. They just never aligned expectations before production.

What does a Melbourne founder need to know about working with offshore QC?

Melbourne has a strong local garment scene, concentrated in the inner northern suburbs like Collingwood, Fitzroy, and Brunswick. These areas still host sample rooms and small-batch producers who can do hands-on QC in person. But most Melbourne brands sourcing at scale work with offshore factories, and their containers arrive through the Port of Melbourne, Australia's busiest container port handling over 3.3 million containers annually. Clothing is one of the port's major containerized imports.

When your shipment lands at Swanson Dock or Webb Dock, defects are already locked in. You cannot rework offshore production after it clears customs. This means QC alignment with your factory matters more than inspection theatrics at the warehouse.

If you are showing at Global Sourcing Expo Melbourne in November 2026 and meeting potential factory partners, bring your defect classification document. Ask them to walk through their inline QC process. A factory that can explain their inspection cadence in detail is a factory that actually runs inspections.

For brands operating in the Melbourne market, the combination of high consumer expectations and distance from production means QC frameworks need to be airtight before the container ships.

Case study: How a Melbourne activewear brand restructured their QC after a failed shipment

A founder running a Melbourne-based performance legging brand came to us after losing an entire 3,000-unit order to a factory in Guangzhou. The shipment failed third-party AQL inspection at the warehouse. Major defects: inconsistent inseam measurements, some pieces 3cm short of spec.

The root cause was not the final inspection. The root cause was cutting. The factory did not verify pattern grading before cutting, and the size 8 pattern was scaled incorrectly. By the time anyone checked measurements, 3,000 leggings were already sewn.

We rebuilt their QC process:

• Pre-production meeting: We review the graded spec sheet together. If measurements look unusual, we flag it before cutting.

• Cut panel audit: After cutting, we measure 10% of panels across sizes before they go to the sewing floor.

• First-off inspection: The first completed unit of each size gets measured and photographed. The founder approves before we continue.

• Inline roving: Inspectors pull random units during production, not just at the end.

• Final AQL at 80% pack: Standard ISO 2859-1 inspection with the founder's specific defect classifications.

Their next order passed AQL with zero major defects. The fix was not tighter final inspection. The fix was catching problems earlier.

What inspection level should you actually specify?

Most brands default to General Inspection Level II. This is appropriate for standard production runs where you have some history with the factory.

Level I uses a smaller sample size and is appropriate when you have established trust and consistent quality history. Level III uses a larger sample and is appropriate for first orders, new styles, or factories with quality problems.

For apparel, the standard configuration is:

• Critical defects: AQL 0 (zero tolerance)
• Major defects: AQL 2.5
• Minor defects: AQL 4.0

Luxury brands often tighten major defects to AQL 1.5. Fast fashion brands sometimes loosen to AQL 4.0 across the board to reduce inspection costs. Neither is wrong. It depends on your customer expectations and margin structure.

"The right AQL depends on your product category and defect risk tolerance. AQL 2.5 is the most common standard for general consumer goods."

What makes a brand easy to work with from the factory QC perspective?

The brands we enjoy working with do three things consistently:

They send complete defect classification documents. Not just "AQL 2.5" in the tech pack. A full list of which issues are critical, which are major, which are minor. With photos if possible.

They approve sealed samples before production. The sealed sample is our reference standard. If you approve a sample with a particular topstitch width, we produce to that width. If you later decide the topstitch is too wide, that is a spec change, not a defect.

They respond to inspection reports promptly. When we flag an issue during inline QC, we need a decision within 24 hours. Continue production? Rework? Adjust tolerance? Silence costs both of us money.

What happens when a lot fails AQL?

A failed lot triggers 100% inspection. We pull every unit, sort defects, and segregate:

• Pass: No defects, ships as-is
• Rework: Fixable defects like loose threads, repairable stitching
• Reject: Unfixable defects like fabric flaws, measurement failures

Rework takes time. A 5,000-unit lot with a 15% defect rate might take a week to sort and repair. Your delivery timeline shifts. If the defect rate is catastrophic, we negotiate. Sometimes the factory absorbs the loss. Sometimes the order gets canceled. Sometimes we produce replacement units at a discounted rate.

The outcome depends on defect cause. If the factory cut the wrong fabric, that is on us. If the brand approved an incorrect spec sheet, that is on them. Most disputes land somewhere in the middle.

Why does AQL inspection matter more in 2026?

Consumer expectations have risen. Social media amplifies quality failures. A TikTok video showing a seam unraveling can cost a brand more than the margin on a thousand units.

At the same time, tariff volatility has tightened margins. Brands cannot afford the cost of returns, refunds, and reputation damage from quality failures. The investment in proper QC pays for itself.

Factories that run rigorous inline QC differentiate themselves. At Ohzehn, we built our quality systems around the principle that final inspection should confirm success, not discover failure. By the time your shipment is cartooned, most problems should already be solved.

The conversation should not be "did the lot pass AQL?" The conversation should be "what did we catch during production and how did we fix it?"

What should your tech pack specify about QC?

Include these elements:

• AQL levels by defect type: Critical 0, Major 2.5, Minor 4.0 (or your variation)
• Inspection level: General II for standard orders
• Defect classification: List of specific issues and their severity
• Measurement tolerances: Which points, what variance is acceptable
• Third-party inspection rights: Name of your inspection company, timing requirements
• Photo documentation: Requirements for inline and final inspection photos

The more specific your tech pack, the fewer arguments later.

AQL is not magic. It is a statistical framework that helps both sides agree on what constitutes acceptable quality. The numbers only work if everyone understands what they mean and aligns before production starts.