AQL 2.5 is a sampling plan, not a quality promise: at General Inspection Level II a lot of 1,200 pieces is inspected by drawing 80 units at random and accepting the shipment if five or fewer major defects are found, rejecting it at six. The plan is published as ISO 2859-1 and it is the commercial standard for soft goods. It is applied alongside a zero tolerance for critical defects and a 4.0 level for minor ones, and its real value is that it puts an objective, repeatable number on a decision that would otherwise be argued about over photographs.
What AQL 2.5 Actually Means
AQL stands for acceptable quality level, and the number is the percentage of defective units a sampling plan is built to tolerate at its boundary. A plan written at 2.5 accepts lots where the defect rate is at or below roughly 2.5 percent with high probability and rejects lots well above it, with a defined risk on both sides. It is a statistical decision rule applied to a random sample, not an inspection of every unit and not a statement that every unit is good.
Two misunderstandings are common enough to be worth killing outright. The first is that AQL 2.5 means 2.5 percent of the shipment is allowed to be defective. It does not; it is the level at which the plan is calibrated, and the accept and reject numbers for a given sample size are what actually govern. The second is that a passed inspection guarantees the lot. It does not; it means a random sample of the stated size met the stated criteria.
| Term | Meaning | Typical value in this category | Who sets it |
|---|---|---|---|
| AQL | Acceptable quality level, the plan's calibration point | 2.5 for major defects | Buyer and programme partner agree |
| Inspection level | How large a sample is drawn relative to lot size | General Inspection Level II | ISO 2859-1 |
| Accept number | Maximum defects allowed in the sample | 5 at a sample of 80 | Read from the standard table |
| Reject number | Defect count at which the lot fails | 6 at a sample of 80 | Read from the standard table |
| Critical defect | Unsafe, unlawful or unusable | Zero tolerance, any count fails | Always zero |
| Minor defect level | Cosmetic faults that do not affect use | 4.0 | Buyer and programme partner agree |
The practical consequence for a buyer is that the plan has to be written down before production, not negotiated after inspection. Sample size, defect definitions and the accept and reject numbers all belong in the order brief, because an inspection carried out against an unwritten plan is an opinion with a report attached to it.
How a Sample Is Drawn
Sample size follows lot size through a code letter table, and the numbers below are the ones that matter for the lot sizes this category actually produces. A 500 piece minimum per colour sits at the small end; a resort or retail programme of 5,000 pieces sits in the middle; a group order of 20,000 sits at the top.
| Lot size, pieces | Code letter | Sample size | Accept at AQL 2.5 | Reject at AQL 2.5 |
|---|---|---|---|---|
| 151 to 280 | G | 32 | 2 | 3 |
| 281 to 500 | H | 50 | 3 | 4 |
| 501 to 1,200 | J | 80 | 5 | 6 |
| 1,201 to 3,200 | K | 125 | 7 | 8 |
| 3,201 to 10,000 | L | 200 | 10 | 11 |
| 10,001 to 35,000 | M | 315 | 14 | 15 |
Two details decide whether the sample means anything. It has to be random across the whole lot rather than drawn from the first cartons packed, because the start of a run and the end of a run behave differently, and a defect that appears only after a tool has been running for hours will be missed by a sample drawn from the front. And cartons have to be selected before the inspector arrives rather than presented to them.
The standard also defines switching rules that most buyers never use and should know about. Consecutive accepted lots can move to a reduced inspection, and rejected lots move to a tightened one with a larger sample and a lower accept number. A programme partner who quotes tightened inspection after a failure is following the standard rather than being difficult.
Sampling procedures for inspection by attributes are published by the International Organization for Standardization as ISO 2859-1, and the numbers above are read from its General Inspection Level II tables. Quoting the standard and the level in an order brief takes one line and removes every argument about sample size.
Defect Classes: Critical, Major and Minor
Defect classification is where most inspection disputes actually start, because the same fault can reasonably be called major or minor depending on how the product is sold. A scratch on a bag headed for a resort amenity programme is a minor cosmetic issue; the same scratch on a bag sold at a premium price point in a boutique is a reject. The classification has to be agreed in advance, in writing, with examples.
| Class | Definition | Typical examples here | Tolerance |
|---|---|---|---|
| Critical | Unsafe, unlawful or unusable | Sharp flash at a handle, a detached handle, a choking hazard on a kids charm | Zero, any count fails the lot |
| Major | Affects use, saleability or life | Crack at a handle root, blocked perforations, a wall below 1.8 mm, a base that will not sit flat | AQL 2.5 |
| Minor | Cosmetic, does not affect use | Light flow lines, a small sink mark, a colour deviation within delta E 3.0 | AQL 4.0 |
| Not a defect | Within specification | A gate mark inside the base, normal parting line, minor tool polish marks | Accepted by definition |
The last row matters as much as the first three. A written specification that also says what is not a defect prevents an inspection from failing on grounds that were never part of the agreement. Gate marks and parting lines are inherent to moulding, and a buyer who rejects them is asking for a process that does not exist.
What Gets Checked on a Moulded Body
The inspection checklist for this product is short because the product is simple, and each item on it corresponds to a fault that actually occurs at scale. The list below covers a standard pre-shipment inspection at final stage.
| Check | Method | Pass criterion | Class if failed |
|---|---|---|---|
| Wall thickness | Calliper at four points on the side wall | Within 0.2 mm of the agreed figure | Major |
| Handle root | Visual plus a 30 cycle flex test on samples | No crack, no whitening at the root | Critical if cracked |
| Perforations | Visual and a gauge on five holes | All open, diameter within 0.5 mm | Major if blocked or undersized |
| Base flatness | Set on a flat surface, press each corner | No rocking beyond 1 mm | Major |
| Flash and sharp edges | Run a cloth along all edges | No snag, no sharp edge anywhere | Critical if sharp |
| Colour consistency | Compare against the approved chip | Within delta E 3.0 | Minor, major if obvious |
| Odour | Smell a sample from the carton | No strong residual odour | Major |
| Carton and packing | Count, weigh and check the packing list | Quantity and markings as agreed | Major |
Flash deserves a note because it is the one fault that can move class quickly. Soft flash at a parting line is cosmetic and usually a minor. Flash that has hardened into a sharp edge at a handle or around a perforation is a safety issue and is classified critical, because it will cut a hand and it will cut it on the first use.
Moulding Faults and How They Are Judged
Moulding faults are specific enough that an experienced inspector recognises them in seconds, and they cluster into a short list. Each has a cause, and each cause has a correction, which is worth knowing because a fault that is identified correctly can be fixed within a run rather than after it.
- Short shot: the body is incomplete at a corner or a handle tip. Caused by insufficient fill or a cold tool. Always a reject.
- Sink mark: a shallow depression opposite a thick section, usually at a handle root. Cosmetic at under 0.3 mm depth, a weakness beyond it.
- Flow line: a visible line where two melt fronts met. Cosmetic unless it sits at a load-bearing point.
- Warp: the body does not sit flat or the rim is not level. Caused by uneven cooling. A major defect beyond about 2 mm.
- Bubble or void: a gas pocket in the wall. Structural if it sits in a load path, cosmetic if it does not.
- Burr at a perforation: a raised ring left by a worn pin. Cosmetic, but it catches towels and is worth rejecting above 0.5 mm.
The judgement that separates a competent inspection from a checkbox exercise is whether a fault sits in a load path. A void in the middle of a side wall is cosmetic; the same void at a handle root is structural. Teaching an inspector to ask that one question is more valuable than adding ten more items to the checklist.
Physical Tests Behind the Numbers
Dimensional and visual checks cover most of what matters, and three physical tests cover the rest. They are run on a small number of units from the sample rather than on all of it, and they are the tests whose results appear in a specification sheet.
| Test | Method reference | Typical criterion | Frequency |
|---|---|---|---|
| Static load test | ASTM physical methods for soft goods | 10 kg for 24 hours, no crack or permanent set | 3 units per lot |
| Loaded drop test | ASTM physical methods | Drop at 0.5 m loaded to 5 kg, no crack | 3 units per lot |
| Handle cycle test | In-house method agreed with the buyer | 500 cycles at 8 kg, no crack at the root | 2 units per lot |
| Durometer hardness | ASTM D2240 | Within 5 points of the agreed figure | 3 units per lot |
| Colour fastness to light | AATCC grey scale method | Grade 4 or better for outdoor colourways | Per colourway, annually |
| Cold flex check | ASTM brittle temperature methods | No crack at minus 20 degrees | Per design, annually |
Physical and mechanical test methods for these properties are published by ASTM International, and quoting the method number rather than the number alone is what makes a result defensible. A drop test result without a drop height, a load figure and a criterion is not a result, it is an anecdote.
Reading the Inspection Report
A pre-shipment inspection report is usually four to eight pages and only a few parts of it decide whether a shipment is released. Read it in this order and the decision takes two minutes rather than twenty.
- Start with the lot size and the sample size, and confirm the sample was drawn to the level you agreed rather than to a smaller one.
- Read the critical defect count first. Any number other than zero ends the discussion regardless of everything else.
- Read the major defect count against the accept and reject numbers for that sample size, not against a percentage.
- Look at photographs of each defect rather than at the description alone, because classification is usually a visual judgement.
- Check the packing section: carton count, quantity per carton, and whether cartons were selected at random.
- Check the conclusion statement. Passed, passed with comments, and failed are three different outcomes and the middle one needs a decision.
The middle outcome is the one that causes problems. A shipment passed with comments is legally released but carries a known issue, and the decision to accept it belongs to the buyer rather than to the inspector. Agreeing in advance what happens in that case, typically a price adjustment or a partial re-run, removes the worst kind of post-shipment argument.
What the Plan Does Not Cover
An AQL plan covers attributes that can be counted in a sample, and there are important things outside it. Knowing the boundary prevents the most painful kind of surprise, which is a shipment that passed inspection and still disappoints.
| Not covered | Why | How to handle it instead |
|---|---|---|
| Performance over time | A sample inspection is a snapshot at one moment | Agree a specification and test it annually |
| Colour drift across batches | Each lot is judged against the chip in isolation | Keep a signed chip and quote the batch code |
| Suitability for a use case | Inspection checks conformance, not design fitness | Test a sample in the real application first |
| Everything outside the sample | Statistical plans accept a defined risk | Raise the level or tighten for high-value lots |
| Packing damage in transit | Inspection happens before loading | Specify packing and photograph the loading |
The third row is the most useful. An inspection confirms that a shipment matches the agreed specification; it does not confirm that the specification was right. A bag that passes inspection at a 1.6 millimetre wall and then creases in a hot resort store room was inspected correctly and specified incorrectly, and no change to the sampling plan would have caught that.
Where Inspection Sits in the Order Cycle
Inspection is one stage in a sequence, and the stages before it are where most defects are actually prevented rather than found. A programme that relies on final inspection alone will reject more lots, ship later, and pay more for both.
| Stage | What is checked | When | Who sees the result |
|---|---|---|---|
| Incoming material check | Compound batch, pigment, any trim | Before production starts | Internal record |
| First article approval | One unit from the start of the run | Start of production | Buyer approves before the run continues |
| In-process check | Wall thickness, colour, flash, cycle stability | Every 2 to 4 hours during the run | Internal record |
| Pre-shipment inspection | Full checklist at the agreed AQL | When packed, before release | Buyer receives the report |
| Loading supervision | Carton count and condition at loading | At the container or truck | Photographs to the buyer |
| Destination check | Quantity and condition on arrival | On receipt | Buyer |
First article approval is the highest-value stage and the one most often skipped under time pressure. Approving one unit from the start of a run costs a day and prevents a 3,000 piece lot being produced to a wrong interpretation of the brief. Skipping it to save that day is the single most expensive shortcut available in a programme of this size.
Writing the Inspection Clause Into an Order
An inspection plan only works if it is in the order document, and the clause is shorter than most people expect. Six lines cover it: the standard, the inspection level, the defect levels, the defect definitions, who pays for the inspection, and what happens on a failure. Anything less and the inspection is a conversation rather than a procedure.
| Clause line | Example wording | Why it is needed |
|---|---|---|
| Standard and level | ISO 2859-1, General Inspection Level II, single sampling | Fixes the sample size |
| Defect levels | Critical 0, major 2.5, minor 4.0 | Fixes the accept and reject numbers |
| Defect definitions | Attached list with photographs | Prevents classification arguments |
| Sample selection | Random across the whole lot, cartons selected by the inspector | Prevents a biased sample |
| Cost | Buyer pays the first inspection, seller pays any re-inspection | Aligns incentives |
| Failure remedy | Re-work and re-inspect, or a price adjustment agreed in writing | Removes the worst argument |
The last two rows are the ones usually omitted and the ones that decide how a failure is resolved. Without a cost allocation there is an argument about who pays for the second inspection, and without a remedy clause there is an argument about whether a failed lot can be sorted and re-presented. Both arguments happen after the shipment is late, which is the worst possible time to be negotiating.
One further line is worth adding for any programme that runs across several seasons: a requirement that the defect definitions and the approved reference chip are re-issued at the start of each season. Specifications drift informally, and a written re-issue once a year costs nothing and prevents the slow accumulation of assumptions that eventually produces a surprising failure.
Programme Notes
Volume programmes and packing specifications are covered in the wholesale EVA tote programme, decoration scope and its effect on inspection are set out in the custom EVA tote guide, and vetting a partner is covered in how to source an EVA tote programme. QUANZHOU JUNYUAN BAGS has run waterproof bag programmes since 2014 with a founding team in this trade since 2004, and inspects every bulk shipment at AQL 2.5 before release.
Monthly output runs at 200,000 units verified by SGS, operations follow BSCI and ISO 9001, materials are screened for REACH and California Prop 65, and every bulk shipment is inspected at AQL 2.5 before release. Programme minimums begin at 500 pieces per design and colour, with sampling at 6 to 10 working days and bulk at 35 to 50 days from approved sample and deposit, shipped FOB Xiamen or DDP to more than 100 countries. Send your lot size, inspection level and defect definitions to service@junyuanbags.com.
People Also Ask
What does AQL 2.5 mean in practice?
A lot of 1,200 pieces is checked by drawing 80 at random, accepting at five or fewer major defects and rejecting at six.
Does AQL 2.5 allow 2.5 percent defective?
No. It is the level the plan is calibrated at. The accept and reject numbers for the sample size are what govern.
What is a critical defect on an EVA tote?
A sharp flash edge, a detached or cracked handle, or a choking hazard on a kids charm. Zero tolerance, any count fails the lot.
How many units are inspected in a 3,000 piece lot?
125 units at General Inspection Level II, with acceptance at seven major defects and rejection at eight.
What does an inspection not cover?
Performance over time, colour drift between batches, and whether the specification was right for the use case.
Why is first article approval worth the delay?
Because it costs one day and prevents a full lot being produced to a wrong interpretation of the brief.
Frequently Asked Questions
Q1. What is AQL 2.5?
An acceptable quality level of 2.5 for major defects, applied through a sampling plan. At a lot of 1,200 pieces the sample is 80 units, accepted at five defects and rejected at six.
Q2. Does it mean 2.5 percent of the shipment may be faulty?
No. It is the calibration point of the plan. The accept and reject numbers for the sample size are what decide the outcome.
Q3. How large is the sample for a 3,000 piece order?
125 units at General Inspection Level II, accepted at seven major defects and rejected at eight.
Q4. What is the standard behind the numbers?
ISO 2859-1, sampling procedures for inspection by attributes, using General Inspection Level II tables.
Q5. What counts as a critical defect?
A sharp flash edge, a cracked or detached handle, and any choking hazard on a children's charm. Zero tolerance applies.
Q6. What is a major defect on this product?
A crack at the handle root, blocked perforations, a wall below the agreed thickness, or a base that will not sit flat.
Q7. What is a minor defect?
Cosmetic faults that do not affect use, such as light flow lines or a small sink mark. Checked at AQL 4.0.
Q8. Should cartons be selected before the inspector arrives?
Yes. Random selection across the whole lot is what makes the sample meaningful, because the start and end of a run differ.
Q9. What physical tests are run alongside the visual check?
A static load test, a loaded drop test, a handle cycle test, durometer hardness and light fastness, each against a named method.
Q10. Does passing inspection guarantee the shipment?
No. It means a random sample of the stated size met the stated criteria, at a defined statistical risk on both sides.
Q11. Why approve a first article before the run continues?
Because it costs one day and stops a full lot being made to a wrong interpretation of the brief, which is the most expensive shortcut in a programme.