Handbag Production Defect Rates: What Is Acceptable at AQL 2.5

What defect rate is acceptable on a handbag order inspected at AQL 2.5? For a 500-piece lot the plan draws 50 units and accepts the shipment on 3 major defects or fewer, rejecting it on 4; for 2,000 pieces it draws 125 and accepts on 7, rejecting on 8. AQL 2.5 is not a promise that 2.5% of your bags may be faulty — it is the process average at which this sampling plan still accepts roughly 19 out of 20 lots, and the same plan will still pass a lot running at 5% defective about three times in four. Read it as a filter with a known curve, not as a quality guarantee, and insist on the written report at AQL 2.5 under ISO 2859-1 with critical defects at zero tolerance.

AQL 2.5 at General Inspection Level II

AQL 2.5 Is a Sampling Plan, Not a Promise About Your Bags

The most expensive misunderstanding in this trade is the buyer who reads AQL 2.5 as two and a half faulty bags in every hundred. It is not that. The acceptance quality limit is defined in ISO 2859-1 as the worst tolerable process average when a continuing series of lots is submitted: it describes the process, not the shipment. The sampling plan built around it has a defined producer's risk of roughly 5%, meaning a process running exactly at 2.5% defective will still have about one lot in twenty rejected. It also has a consumer's side, which is the part buyers never see: the same plan is far more permissive than 2.5% suggests once the process drifts.

A Brisbane accessories brand learned this on a 2,000-piece order of structured satchels. The line was running well, the pre-shipment inspection at AQL 2.5 drew 125 units and found 6 majors, which passed. The goods shipped, and eight weeks later the marketplace return rate on that SKU came back at 3.1% against a category norm nearer 1%. Nothing had been mis-inspected: 6 defects in 125 is a pass, and a lot with about 5% real defects had roughly a three-in-four chance of passing this plan. The number was never a ceiling on outgoing quality — it was a decision rule on a sample.

Accept and Reject Numbers Across Five Lot Bands

These are the numbers that matter on a real purchase order. They come from ISO 2859-1 at General Inspection Level II, single sampling, normal inspection, with AQL 2.5 applied to the major class. Notice that the sample grows far more slowly than the lot: coverage is 10% at 500 pieces and 4% at 5,000, which is the compromise the standard makes and the reason inspection is a filter rather than a sieve.

Two consequences follow directly. First, a 500-piece first order is inspected at ten per cent coverage, which is generous, and the accept number of 3 is tight relative to the draw; small lots are genuinely harder to pass than buyers assume. Second, as the programme grows the percentage inspected falls, so a brand scaling from 500 to 5,000 pieces should not read a continued pass record as continued control. Where the volume justifies it, hold the standard and add an in-process check rather than relying on the pre-shipment draw alone.

Severity Classes and the Zero Tolerance Tier

Every defect found is assigned a class before it is counted, and the class decides both the number and the consequence. Three tiers are standard practice, and the top tier has no arithmetic at all.

• Critical — accept 0, reject 1. Anything that can injure a user, breach a regulation or make the bag unsellable on safety grounds: a sharp exposed edge on hardware, a needle fragment left in the lining, a detachable small part on a children's item, or a restricted substance finding under REACH Annex XVII or Proposition 65.
• Major — AQL 2.5. A fault a customer would notice and return the bag for: a skipped stitch run, a strap attachment that fails the pull test, a zipper that jams, a logo positioned outside tolerance, or shade variation beyond the agreed Delta E.
• Minor — commonly AQL 4.0. A fault a professional eye notices but a customer would accept: a thread tail under 3 mm, a light crease that relaxes on hanging, a carton mark printed slightly off register.

Running critical at zero and minors at AQL 4.0 on the same draw is the most common professional configuration, and it is worth stating explicitly in the purchase order rather than leaving the classes to the inspector's judgement on the day. Where a brand sells into a marketplace with a strict return policy, some buyers tighten majors to AQL 1.0; the cost of that decision is more rejected lots, not better bags, because the sampling plan does not change what comes off the line.

AQL 2.5 Measured Against 1.0, 4.0 and Zero Tolerance

Choosing an AQL is choosing where to stand on the operating characteristic curve. The comparison below uses the same 2,000-piece lot so the trade-off is visible: a tighter number rejects more lots at the same underlying quality, and a looser one ships more.

Zero-acceptance looks rigorous and behaves badly. At a true defect rate of 2.5%, a plan that accepts on zero defects out of 125 will reject roughly 96% of lots, which means every shipment becomes a re-sort, the calendar absorbs weeks and the conversation turns adversarial without a single better bag being made. AQL 2.5 sits at the point where a controlled process is not punished and a drifting one is caught often enough to matter. It is the standard written into programmes run through our validated partner factory Quanzhou Yingming Bags, and it is the one most retail buyers recognise without explanation.

Six Defect Measurements an Inspector Takes First

Defect classification is only as good as the measurements underneath it. Six checks produce most of the findings on a handbag inspection, and each one needs a number in the specification or the inspector is left guessing.

1. Stitch density and continuity. 7-9 stitches per inch with Tex 40 thread and a 10 mm seam allowance; two or more skipped stitches inside any 10 cm run is a major.
2. Attachment strength. A strap or handle attachment pulled at 20 kg for 10 seconds with no stitch failure or tearing at the joint; this is the failure that generates field returns.
3. Dimensional conformance. Body width, height and gusset within plus or minus 3 mm of the tech pack, strap drop within plus or minus 5 mm.
4. Colour conformance. Delta E 2.0 or better against the approved physical swatch under a defined light source, with shade variation between panels in one bag judged as a separate major.
5. Hardware condition. Plating thickness in the 8-12 micron range with 24 hours to first corrosion tested to ASTM B117; visible plating faults, sharp edges or a closure that does not seat flush are majors.
6. Zipper endurance. On a #5 metal or nylon chain, 500 open-and-close cycles without tooth loss, slider deformation or tape fraying.

Normal, Tightened and Reduced: the Switching Sequence

ISO 2859-1 is not a single table; it is three tables with rules for moving between them. The switching rules are what turn a series of inspections into a system, and they are the part of the standard almost nobody reads until a dispute.

• Normal to tightened when 2 out of 5 consecutive lots, or fewer, are rejected on original inspection. Tightened inspection draws the same sample size but applies a stricter accept number, so the same line suddenly fails more often.
• Tightened back to normal after 5 consecutive lots are accepted under tightened inspection. This is the escape route, and it requires five clean runs in a row.
• Normal to reduced after 10 consecutive accepted lots with production at a steady rate and approval from the responsible authority. Reduced inspection draws fewer units and is a reward, not a right.
• Discontinuation where 5 consecutive lots are rejected while on tightened inspection. At that point the standard says stop inspecting and fix the process, because inspection is no longer the constraint.

For a brand, the practical reading is simple: ask which state the programme is currently in. A sourcing partner who can answer that question is keeping records across orders; one who cannot is treating each inspection as an isolated event, which is precisely the situation the switching rules were written to prevent.

The Four Things an Accept Number Hides

A pass at AQL 2.5 is worth having and it is not the whole picture. Four things sit outside the accept number, and each of them has cost a brand money.

• Defects outside the sample. With 125 units drawn from 2,000, roughly 94% of the shipment is never opened. Any fault that is systematic but below the sampling threshold ships in full.
• Defects the checklist does not cover. Inspection follows an agreed checklist. Comfort, hand feel, how the bag sits on a shoulder and how the lining behaves after a week are not on it.
• Drift between approved sample and bulk. A bag can meet every written tolerance and still not feel like the approved sample, because the specification captured dimensions and not the hand feel; the fix is a sealed golden sample at the line, not a tighter AQL.
• Latent failures. Adhesive creep, plating tarnish and edge-paint cracking appear after weeks of use, long after the inspection report is filed.

The sensible response is not to tighten the AQL but to add two things: a pilot run on the line before bulk starts, and a retained golden sample that the line is measured against. Both are cheaper than a rejected shipment, and both attack the defect rate at the point where it is set rather than at the point where it is counted. The cost drivers behind a quotation are relevant here too, because stitch density and plating hours are the two lines most often cut to hit a price.

The Inspection Clause, Line by Line

Six lines turn an intention into an enforceable clause. Copy them into the purchase order and the inspection stops being a conversation.

1. Standard and level. Pre-shipment inspection to ISO 2859-1, General Inspection Level II, single sampling, normal inspection.
2. AQL by class. Critical 0, major 2.5, minor 4.0, with the classes defined in the specification rather than left open.
3. Timing. Inspection after packing is complete and before the container is loaded, with the buyer given 5 working days' notice.
4. Report and evidence. A written report with the lot size, sample size, defect list by class and the pass or fail decision, plus photographs of every major and critical finding.
5. Failure remedy. Re-sort or rework at the seller's cost, re-inspection at the seller's cost, and a defined latest shipment date that does not move because the first inspection failed.
6. Third-party option. The buyer may appoint SGS or another nominated inspector, with access to the packed cartons and the production records.

Reading an Inspection Report in Four Minutes

Reports are long and most of the length is administrative. Four checks extract everything that matters: confirm the lot size and the sample size agree with the standard for that band; read the critical line first and stop if it is not zero; group the majors by family rather than by count, because six majors of the same type is a process fault and six unrelated majors is noise; and compare the defect list against the golden sample notes to see whether a fault was already visible at approval. A report that only gives a pass or fail with no defect breakdown is not a report, it is an opinion, and it should be sent back the same day.

Where the report keeps producing majors in the same family, the remedy is upstream: the stage-by-stage manufacturing process shows where each defect family originates, and the twelve stages from cut to carton identify which four of them slip most often. Fixing the stage is cheaper than tightening the standard, and it is the only route to a defect rate that improves across orders rather than being re-measured each time.

Where Defects Are Made: A Pareto by Operation

An accept number tells a buyer how many faults were found; it does not say which operation made them. Grouping defects by the stage that produced them is the fastest route to a lower rate, because the distribution is stable across most programmes. Stitching produces the largest share, assembly the second and finishing the third, while cutting and skiving together account for well under a fifth. Fixing the top two families moves the rate further than tightening any AQL.

The distribution is useful because it is counter-intuitive. The operations that look most skilled, cutting and skiving, produce the fewest majors; the ones that look routine, closing and assembly, produce two thirds of them. A corrective note aimed at the top two families is worth more than a re-sort, and it is the only route to a rate that improves across orders rather than being re-measured each time.

How Much One Inspector Gets Through in a Day

A pre-shipment inspection is bought by the man-day, and knowing what fits in one changes how the booking is specified. A trained inspector working a handbag with roughly 40 measurement and function points completes 12-20 units an hour, so an 80-unit draw takes five to seven hours and a 125-unit draw eight to ten, before the report is written. Two inspectors halve the clock and do not halve the findings.

• 12-20 units per hour at about 40 check points per unit, including a photograph of every major.
• 80 units: 5-7 hours. 125 units: 8-10 hours. 200 units needs a second day or a second inspector.
• Add an hour per extra colourway, because each is treated as its own lot with its own measurement record.
• Book at 80% packed, or the packing method and the carton marks cannot be checked at all.

The consequence for a buyer is that a two-man-day booking is not a luxury above about 2,000 pieces; it is what lets the draw stay at the standard sample size rather than being quietly reduced to fit the day. An inspection that ran out of time is an inspection that drew fewer units than the plan requires, and nothing on the report will say so.

Frequently Asked Questions

People Also Ask