AQL Inspection Levels for Luxury Bags Compared
Acceptable Quality Limit is a threshold, not a promise. On a luxury handbag order it sets the largest number of minor, major and critical faults an inspector may accept in a random sample before the lot is refused. The inspection level you name — I, II or III — decides how many pieces are pulled off the bench, and that sample size decides how much risk reaches the container. Most luxury programmes run AQL 2.5 at general inspection level II, which on a 500-piece order means 50 pieces examined and an accept number of 3 against a reject number of 4 for major defects. This guide compares the levels, shows the sample size maths behind each one, and explains where a tighter or looser setting costs you money.
What AQL actually means for a luxury handbag order
AQL is the worst average quality a shipment may carry and still pass. It is expressed as a percentage, or as defects per hundred units, and it is always read together with a sample size code letter. The Acceptable Quality Limit does not promise that a lot is flawless; it caps the number of faults an inspector will tolerate in a random draw. On luxury programmes we treat it as a floor rather than a target, because a two per cent tolerance on a hand-set handle means a visible failure on the piece a client lifts out of the dust bag.
Three variables govern every inspection: the lot size, the inspection level, and the AQL itself. Change any one of them and the accept and reject numbers move with it. A brand that specifies only the AQL and leaves the inspection level undefined has handed back the most important lever it owns, because level I can pass a lot that level III would refuse on precisely the same defect count.
| Lot size (pcs) | Level I code / pieces | Level II code / pieces | Level III code / pieces | AQL 2.5 accept / reject at level II |
|---|---|---|---|---|
| 51–90 | D / 8 | E / 13 | F / 20 | 1 / 2 |
| 91–150 | E / 13 | F / 20 | G / 32 | 1 / 2 |
| 151–280 | F / 20 | G / 32 | H / 50 | 2 / 3 |
| 281–500 | G / 32 | H / 50 | J / 80 | 3 / 4 |
| 501–1200 | H / 50 | J / 80 | K / 125 | 5 / 6 |
| 1201–3200 | J / 80 | K / 125 | L / 200 | 7 / 8 |
Read the 281 to 500 row first. It covers the most common opening order size in our programmes, and it is where the difference between levels becomes concrete: level I draws 32 pieces and tolerates two major faults, level II draws 50 and tolerates three, level III draws 80 and tolerates five. The stricter setting is not the one with the smallest accept number. It is the one that puts enough pieces under the lamp for the fault to surface at all.
Inspection levels I, II and III compared
General inspection levels differ only in the relationship between lot size and sample size. Level I is the least protective: it reads one code letter below level II and therefore pulls roughly a third fewer pieces. Level III is the most protective, reading one code letter above. The AQL stays the same across all three; what changes is the statistical power of the draw. A level I inspection of a 500-piece lot examines 32 units, so a fault running at two per cent has a meaningful chance of not appearing in the sample at all.
- Level I suits low-risk repeat styles where the last five lots passed cleanly and the construction has not changed.
- Level II is the default for first orders, new colourways, new leather lots and anything carrying plated hardware.
- Level III is appropriate for launch pieces, runway-linked styles, and any order where a recall would damage the maison's name.
- All three levels use the same defect classification, so a major fault is a major fault regardless of how many pieces were drawn.
- Inspection cost scales with sample size: moving a 500-piece lot from level I to level III roughly doubles bench hours for that lot.
- Switching between levels is a commercial decision, but it should be recorded in the order file so the next lot is audited on the same basis.
Why level II is the default on bag programmes
Level II is the equilibrium point between detection power and bench time. On a 500-piece lot, 50 pieces is enough that a two per cent defect rate will usually show itself, and it is small enough that a full inspection — external surfaces, lining, hardware cycles, stitch density, edge finishing, measurement against the approved sample — can be completed inside a single working session. In our atelier a level II draw of 50 structured bags takes about two and a half working hours on the bench, including the written report and photographic evidence for each finding.
General versus special inspection levels
ISO 2859-1 also defines four special inspection levels, S-1 through S-4, which sit below the general levels and are reserved for tests that destroy the sample or consume measurable labour. On a 500-piece lot these levels draw somewhere between three and thirteen pieces rather than fifty. They exist because some checks cannot be run on fifty units: a destructive pull test on a handle attachment, a salt spray cycle on a plated clasp, or a rub fastness run on a dyed suede panel all consume the piece they are performed on.
Where special levels earn their place
We pair the two systems rather than choosing between them. The visual and dimensional inspection runs at general level II on 50 pieces. The destructive and instrument-based checks run at a special level on a handful of pieces, usually drawn from the same randomised set so the results can be read together. This keeps the expensive tests within reach while the large sample still carries the statistical weight of the decision.
- Handle and strap attachment pull testing, destructive by nature, run at a special level.
- Salt spray exposure on plated hardware, which consumes the finish being tested.
- Colour fastness to rubbing on dyed suede and vegetable-tanned panels.
- Edge paint adhesion cycling, where the coating is abraded to failure.
- Lining seam strength measurement on woven reinforcements.
Critical, major and minor defects in bag terms
A defect class is worth more than the AQL number. Most programmes set critical defects at zero tolerance regardless of sample size, because a single critical fault — a sharp edge, a detachable small part, a needle fragment left in the structure — is not a statistical question. It is a safety and reputation question, and one occurrence is already one too many.
| Class | Definition on a bag programme | Typical AQL applied | Examples |
|---|---|---|---|
| Critical | Renders the piece unsafe, illegal or unsaleable in any market | 0 — one found, lot refused | Broken needle fragment, sharp hardware burr, detachable decorative stud |
| Major | Visible or functional fault a client would reject on sight | 2.5 | Crooked topstitch, colour mismatch between panels, stiff turn-lock, loose handle rivet |
| Minor | Deviation a trained eye notices but a retail client would likely accept | 4.0 | Slight edge paint sheen variance, a single skipped stitch inside the lining, dust bag creasing |
Critical
Zero tolerance. One occurrence in the sample stops the shipment, triggers a full sort and a written root cause note before anything moves.
Major
A client-facing fault. At AQL 2.5 on 50 pieces the lot is accepted at three and refused at four, and a refusal moves the next lot to tightened inspection.
Minor
A finishing deviation recorded for the record. At AQL 4.0 on 50 pieces the tolerance is five accepted against six refused, and repeated minors signal a bench skill gap.
Borderline
Any fault the inspector cannot classify alone is escalated to the atelier lead and judged against the approved sample, not against opinion.
AQL 2.5 against 1.5 and 4.0
The AQL column changes the accept and reject pair for a fixed sample size. Holding the draw at 50 pieces on a 500-piece lot, AQL 1.5 accepts two and refuses three, AQL 2.5 accepts three and refuses four, and AQL 4.0 accepts five and refuses six. The spread looks small on paper and is large in practice: the difference between AQL 1.5 and AQL 4.0 on the same 500-piece lot is roughly three times the tolerance for faults a client will see immediately.
| AQL | Accept / reject at 50 pieces | What it tolerates on a 500-piece lot | Where it fits |
|---|---|---|---|
| 1.5 | 2 / 3 | Around six to eight visible faults before refusal | Runway pieces, exotic skins, hand-set stone or resin work |
| 2.5 | 3 / 4 | Around eight to twelve visible faults before refusal | Standard for structured calfskin and most seasonal programmes |
| 4.0 | 5 / 6 | Around fifteen to twenty visible faults before refusal | Simple constructions, canvas bodies, unlined totes |
We quote AQL 2.5 as the standing setting for major defects because it matches what a luxury client actually rejects. Below 1.5 the cost of inspection and rework rises faster than the perceived quality gain; above 2.5 the number of visible faults reaching a boutique floor becomes a returns problem rather than a quality control question. Where a brand asks for something tighter, we apply AQL 1.5 to major defects and keep critical defects at zero, which is the configuration most maison quality teams specify.
Switching rules: normal, tightened and reduced
ISO 2859-1 is not a static table. It includes switching rules that move a programme between normal, tightened and reduced inspection based on recent history, and those rules are what turn a single inspection into a system. Tightened inspection raises the bar after failure; reduced inspection lowers it after sustained success. Both are automatic in a well-run programme, and both should be documented rather than improvised.
- Start at normal inspection with the agreed level and AQL, recorded on the order file before the first piece is cut.
- Move to tightened inspection when two out of five consecutive lots are refused, which lowers the accept number and raises sample discipline.
- Return to normal inspection once five consecutive lots pass under tightened rules, with the movement dated and signed off.
- Qualify for reduced inspection only after ten consecutive passing lots, and only if production has been steady and no construction change is pending.
- Suspend reduced inspection immediately and return to normal the moment any lot is refused or a material or tooling change is introduced.
Reading accept and reject numbers without confusion
The accept number is the maximum count of defects in the sample that still passes the lot. The reject number is one above it, and there is no middle ground. At AQL 2.5 on 50 pieces, three major faults pass and four refuse. A common error is to read the AQL percentage as a promise about the whole lot: it is not. It describes the sampling plan's tolerance, not a guarantee that the remaining 450 pieces are clean. The correct reading is that the plan will usually accept lots at or better than the stated quality level, and will increasingly refuse lots worse than it.
Sample size decides whether a fault can be found; the AQL only decides how many you tolerate once it is found. Specify both, in writing, before cutting begins.
Where inspection level meets real risk
Two failure modes dominate luxury bag programmes, and both are invisible to a small sample. The first is a material drift: a new dye lot of calfskin that sits half a shade cooler than the approved reference, or a plating bath that runs a micron thinner than the one that produced the sample. The second is a bench drift: a new artisan on the edge paint station whose coating is a fraction heavier, or a machine whose stitch tension has slipped out of tolerance over a fortnight. Both show as a pattern spread across hundreds of pieces, and both are caught by measurement and comparison rather than by a quick glance.
- Standing AQL for major defects
- 2.5, ISO 2859-1, single sampling, normal inspection
- Critical defects
- Zero tolerance, one occurrence refuses the lot
- Minor defects
- AQL 4.0 recorded separately from major findings
- Default inspection level
- General level II, 50 pieces on a 500-piece lot
- Bench time for a level II draw
- About 2.5 working hours including report and photographic evidence
- Panel alignment tolerance
- ±0.5 mm against the approved sample
- Topstitch density reference
- 8 stitches per inch on structured calfskin
- Hardware cycling
- 20 open-and-close cycles per closure on every drawn piece
- Minimum order quantity
- 500 pieces per style per colourway
- Sampling lead time
- 6 to 10 working days, 12 to 15 for complex constructions
- Bulk lead time
- 35 to 50 days from approved sample
- Shipment terms
- FOB Xiamen
- Programme history
- Export work since 2014 across 7 production lines
- Monthly capacity
- About 200,000 pieces per month
The cost of choosing the wrong inspection level
Choosing level I to save bench hours is the single most expensive economy available on a bag programme. A level I draw of 32 pieces on a 500-piece lot will pass on two major faults, which sounds strict until you consider that the remaining 468 pieces were never examined. If the underlying fault rate is three per cent, that lot is statistically likely to ship with well over a dozen client-visible defects, and the brand absorbs them as boutique returns, courier costs and a damaged relationship with the buyer who placed the piece on the floor.
What a single escaped fault costs
The arithmetic is unflattering. Reworking one closure on a returned piece costs around USD 4.80 in parts and bench time, before any freight is counted. Two return courier legs and a client service touch on the same piece routinely cost several times that figure, and the reputational cost of a refused re-order is larger again. Against that, the incremental bench time of moving from level I to level II on a 500-piece lot is a fraction of a working day. The inspection level is not where savings should be found.
How we run level II on a 500-piece run
- Confirm the lot size, the inspection level and the AQL in writing, and confirm which sample is the approved reference for comparison.
- Randomise the draw across cartons and production lines so the sample is not skewed toward one bench or one shift.
- Inspect external surfaces first: leather grain, shade continuity, panel alignment, topstitch line, edge paint continuity and hardware seating.
- Move to function: cycle every closure twenty times, test handle attachment by hand and by instrument, and open and close every lining pocket.
- Measure against the approved sample and record deviations greater than 0.5 mm, photographing each one with a scale reference.
- Classify every finding as critical, major or minor, apply the accept and reject pair, and issue the report with photographs before the lot is released.
The comparison in one paragraph
- Inspection level sets sample size, the AQL sets tolerance, and the defect class sets consequence. On a 500-piece luxury bag lot the standard configuration is general level II with AQL 2.5 for major defects and zero tolerance for critical defects, which means 50 pieces are examined and the lot passes at three major faults and is refused at four. Level I saves bench hours and buys risk; level III buys assurance and costs time. Programmes that hold level II, document switching rules and compare every piece against an approved sample see fewer escaped faults than programmes that argue about the AQL number alone.
Set the inspection level before you cut
Send us your style, colourway and order size and we will specify the sampling plan, the defect classification and the accept and reject numbers in writing before the first hide is cut. Request a quality consultation and quotation at /contact-quote/.
Frequently Asked Questions
What is the difference between an AQL and an inspection level?
The AQL sets how many defects are tolerated in the sample; the inspection level sets how many pieces are drawn. Level I pulls fewer pieces, level III pulls more, and both use the same AQL table. Specify both in writing before cutting begins.
Why is level II the default for luxury handbag orders?
Level II balances detection power against bench time. On a 500-piece lot it draws 50 pieces, enough for a two per cent fault rate to surface, while still allowing a full external, functional and dimensional inspection inside one working session.
How many pieces are inspected on a 500-piece order at level II?
Fifty pieces, corresponding to sample size code letter H under ISO 2859-1. At AQL 2.5 for major defects the lot is accepted at three faults and refused at four. Critical defects carry zero tolerance at any sample size.
Does a lower AQL always mean better quality?
Not by itself. AQL 1.5 on a small sample can be less protective than AQL 2.5 on a larger one, because the larger draw is more likely to find the fault. Sample size and tolerance have to be specified together to mean anything.
What are special inspection levels used for?
Special levels S-1 through S-4 are reserved for tests that destroy or consume the sample, such as handle pull testing, salt spray on plated hardware and rub fastness on dyed suede. They run alongside the general level draw, not instead of it.
What happens if a lot is refused at final inspection?
The lot is held, a full sort is performed, and the defects are traced to a root cause before re-presentation. Two refusals inside five consecutive lots move the programme to tightened inspection with a lower accept number.
Should critical defects ever have a tolerance above zero?
No. Critical defects are those that make a piece unsafe, illegal or unsaleable, such as a needle fragment or a sharp hardware burr. One occurrence refuses the lot and triggers a documented root cause investigation before anything ships.
How long does a level II inspection take?
On a 500-piece structured bag lot, examining 50 pieces including external surfaces, hardware cycling, measurement and the written report with photographic evidence takes about two and a half working hours on the bench.
Can the inspection level change between orders?
Yes, and the switching rules in ISO 2859-1 govern how. A programme can move to tightened inspection after repeated refusals and to reduced inspection after ten consecutive passing lots, provided production is steady and no construction change is pending.
What tolerance should I set for panel alignment?
We work to ±0.5 mm against the approved sample for structured calfskin bodies. That figure is tight enough to be invisible to a client and loose enough not to generate rejections for variation no one can see.
How is the sample randomised fairly?
Cartons are drawn from across the lot rather than from the top layer, spanning production lines and shifts, so the sample is not skewed toward one bench or one artisan. The selection is recorded alongside the inspection report.
What does it cost to rework a returned piece?
Reworking one closure on a returned piece costs around USD 4.80 in parts and bench time, before freight or client service handling. Two courier legs and the associated service time routinely cost several times that amount.
Is AQL 2.5 written into every order?
AQL 2.5 for major defects is our standing setting, with AQL 4.0 for minor defects recorded separately and zero tolerance for critical defects. A brand may specify AQL 1.5 instead, which we apply to major defects and note on the order file.
Does the inspection level affect the price of the order?
Inspection is included in the programme, so changing the level changes bench hours rather than the unit price. Moving a 500-piece lot from level I to level III roughly doubles the inspection time for that lot.
People Also Ask
Which AQL level is best for luxury handbags?
AQL 2.5 at general inspection level II is the standing choice for structured calfskin programmes. Exotic skins and hand-set stone work justify AQL 1.5, while simple canvas or unlined constructions can run at AQL 4.0.
How do I read an AQL table?
Find your lot size in the left column, read across to your inspection level for the code letter, then look up that code against your AQL to get the sample size and the accept and reject pair.
What is general inspection level II?
It is the middle of three general levels defined in ISO 2859-1. It maps a 281 to 500 piece lot to code letter H and a sample of 50 pieces, and it is the default starting point for most programmes.
What is a critical defect in a handbag?
A critical defect makes the piece unsafe or unsaleable, such as a broken needle fragment left in the structure or a sharp burr on plated hardware. It carries zero tolerance regardless of sample size or AQL.
Can I use AQL for destructive testing?
Yes, but through the special levels S-1 to S-4, which draw far fewer pieces because the test destroys or consumes the sample. Destructive checks run alongside the general level draw rather than replacing it.
Sources and further reading
Every figure on this page is either our own programme data or attributed below. Open the originals before you write them into a specification.
- ISO 2859-1:1999 Sampling procedures for inspection by attributesInternational Organization for Standardization Defines the sample size code letters, general and special inspection levels, and the accept-reject tables used throughout this guide.
- ISO 2859 series on acceptance sampling by attributesInternational Organization for Standardization Companion parts covering skip-lot and sequential sampling used when a programme qualifies for reduced inspection.
- Leather Working Group environmental and traceability audit protocolsLeather Working Group Referenced for leather traceability expectations that sit alongside dimensional and visual inspection on luxury programmes.
- Consumer product safety and recall guidanceU.S. Consumer Product Safety Commission Basis for treating critical defects such as detachable small parts and sharp hardware edges at zero tolerance.
- Reporting on quality, compliance and sourcing in luxury supply chainsSourcing Journal Industry reporting on why inspection depth, rather than headline AQL figures, drives escaped defect rates.