AQL 2.5 Inspection for Golf Bags Explained for Buyers

AQL 2.5 under ISO 2859-1 fixes two numbers for a shipment: how many units to pull out of the lot, and how many of those may be defective before the whole lot is refused. It is a sampling scheme, not a guarantee, and the difference matters commercially. At 2.5 per cent, a lot that is genuinely running at 2.5 per cent defective still stands roughly a 95 per cent chance of being accepted, which is the point of the scheme and also the reason it cannot be the only quality control a buyer relies on. This walkthrough covers the sampling table for bag lot sizes, how to classify defects so the numbers mean something, what an inspection day actually looks like, and what to do when a lot fails.

Related reading: Golf Bag Factory Audit Checklist for Importers

What AQL 2.5 Actually Commits You To

AQL stands for acceptance quality limit, and the number after it describes the worst tolerable process average that a sampling plan will still accept most of the time. Choosing 2.5 does not mean the supplier may ship 2.5 per cent defective bags, and it does not mean 2.5 per cent of a lot will be inspected. It means the plan is calibrated so that lots at or better than 2.5 per cent defective are accepted with high probability, and lots substantially worse than that are rejected with increasing frequency.

The standard behind the number is ISO 2859-1, the attribute sampling scheme derived from the older MIL-STD-105 lineage. It defines three inspection levels, general I, II and III, plus four special levels. General level II is the default for consumer goods and the level used on the programmes we run, because level I samples too thinly for a stitched product and level III costs more inspection time than the extra confidence is worth on a repeat order.

Two properties of the scheme are worth understanding before a purchase order is written. First, the sample size is driven by lot size and inspection level, not by AQL; the AQL only sets the accept and reject numbers for that sample. Second, the protection is asymmetric. The producer risk of having a good lot rejected is low, the consumer risk of accepting a bad lot falls as the defect rate rises above the AQL, and neither is zero at any quantity.

That asymmetry is why a strict buyer should not simply lower the AQL number. Moving from 2.5 to 1.5 on a 500-piece lot changes the sample from 50 to 50 and the acceptance number from 3 to 2, which is a real tightening, but it does not close the gap that sampling always leaves. What closes the gap is process control at the source, which is a different instrument: a pre-production meeting, an inline check at the first 50 pieces, and a final inspection as a verification rather than as a filter.

Reading the Sampling Table for Bag Lot Sizes

The table below is what an inspector works from when the plan is general level II and the quality limit is 2.5. Lot size picks the code letter, the code letter picks the sample size, and the AQL picks the acceptance and rejection numbers. Read across a row, not down a column.

A 200-piece order therefore sits in the 151-280 band, which gives code letter F, a sample of 32 bags, an acceptance number of two and a rejection number of three. A 600-piece order sits in H: 80 bags pulled, five majors accepted, six rejecting the lot. Note that the sample grows far more slowly than the lot. Going from 200 to 600 pieces adds 48 sampled units, not 400, which is the efficiency that makes the scheme usable and the blind spot that makes it incomplete.

The sample must be drawn randomly across the lot, not from the top carton. In practice that means pulling cartons at intervals through the stack, taking units from the middle as well as the ends, and recording which carton each unit came from so a defect cluster can be traced back. A sample pulled entirely from the last three cartons inspected a line that had already been corrected, or a line that was already drifting, depending on the day.

One practical note on lot definition: the lot should be one style, one colourway and one production run. Combining two colourways into a single 600-piece lot halves the inspection cost and doubles the chance that a problem in one colour hides inside an acceptable average. If a programme has three colourways, inspect three lots, or accept that the report describes the average rather than any single colour.

Critical, Major and Minor: Writing the Defect List

The sampling table counts defective units, and a unit is defective if it fails any classification rule. That makes the defect list the most consequential page in the inspection package, because it decides what gets counted. A list that classifies everything as major turns a cosmetic thread end into a rejection; a list that classifies everything as minor turns a broken zip into an acceptance.

The working convention on bag programmes is three classes. Critical covers safety, legality and anything that makes the product unsaleable: a needle fragment, mould, a sharp exposed moulding edge, a wrong country-of-origin label, a missing warning label where the destination requires one. Critical is normally zero acceptance regardless of AQL. Major covers function and obvious visual failure at retail: a seam that has opened, a zip that will not run, a hole, a strap anchor with no bar-tack, a colour mismatch visible at arm length, a logo in the wrong position. Minor covers cosmetic deviations inside tolerance: a loose thread under 20 mm, chalk residue, uneven topstitch under 2 mm, a light scuff on a base that will not be seen.

Two rules that keep the list honest

Two rules keep the list honest. Every line should reference where the requirement comes from: a callout number, a point of measure, a BOM line, a legal requirement. And every line should be decidable by two people looking at the same bag. A defect definition that requires judgement about intent cannot be counted, and one that cannot be counted cannot be compared between inspections.

The list also needs a rule for accumulating minors. Most schemes fail a lot when either the major count exceeds Ac or the total defect count reaches a stated ceiling, often 2.5 times the major acceptance number. Without that ceiling, a lot with twelve minor defects and five majors sits exactly on the edge of acceptance and the argument is about arithmetic rather than about the bags.

Inside the Inspection Day

A pre-shipment inspection on a bag order takes one man-day for most lots up to 1,200 pieces, and two for larger or more complex programmes. The day has a fixed shape, and knowing it helps a buyer judge whether the report is credible.

1. Lot identification. The inspector confirms the quantity on the floor matches the packing list, and confirms the style, colourway and revision against the approved pack.
2. Sample drawing. Units are pulled at intervals through the cartons and logged by carton number, so any cluster can be traced to a point in the line.
3. Reference check. The approved sample and the measurement chart are laid out before anything is measured. Inspection without the approved sample on the table is opinion, not inspection.
4. Dimensional measurement. Every point of measure on the chart is measured on every sampled unit, with a steel rule or a gauge, and recorded per unit rather than as an average.
5. Functional checks. Zips run the full length, straps loaded and released, legs deployed and retracted on a stand bag, cart strap threaded, dividers checked with a gauge, hood fitted.
6. Visual examination. Seams, stitch density, thread ends, colour against the standard, logo position and artwork quality, hardware finish, cleanliness inside and out.
7. Classification and counting. Each defective unit is classified against the list, and the counts are compared with the acceptance and rejection numbers.
8. Report and photographs. A written report with the counts, the measurements, defect photographs and a clear pass or fail verdict, issued the same or next day.

The order matters. Measuring before checking the reference produces numbers that cannot be judged, and classifying before counting produces a verdict that drifts as the inspector gets tired. On a 125-unit sample with 30 minutes of measurement per unit, fatigue is a real variable and the sequence is what keeps it from changing the outcome.

The single most common weakness in a report is an average measurement. Reporting that the top oval averaged 254 mm tells the buyer nothing about the unit that measured 248 mm. Ask for the per-unit figures, or at minimum the minimum and maximum on each critical dimension, and check the spread rather than the mean.

The Measurements That Fail Lots

Across bag programmes, dimensional failures cluster on a handful of points rather than spreading evenly. Knowing which ones to watch lets a buyer put the inspection effort where the risk is.

• Top oval, both axes. The most frequent dimensional failure, because it is affected by the collar, the divider, the lining and the assembly sequence. Tolerance is normally plus or minus 3 mm.
• Finished weight. One number that catches a swapped shell, a reinforcement left out or a padding layer that does not belong. Tolerance is typically plus or minus 50 g on a 2.35 kg bag.
• Strap anchor position and stitch. Anchor geometry drives carry balance, and a missing bar-tack is a functional failure that no dimensional check will catch.
• Logo position. A 2-3 mm drift is invisible in isolation and obvious against the approved sample, which is why it needs a dimensioned placement from a seam.
• Divider wall height and gauge fit. A well that measures correctly but binds on a 45 mm gauge will fail in the customer hands, so the functional check matters as much as the measurement.
• Colour against the standard. dE 1.5 for shell, dE 1.0 within a lot, measured under D65 rather than in the warehouse doorway.

A second cluster is construction rather than dimension: binding width, pocket symmetry and the finish at the pocket corners. These are the defects that a buyer sees in a photograph and an inspector classifies as minor, which is exactly the mismatch that makes a passed lot feel like a failed one. If pocket symmetry matters, it has to be in the defect list as a major with a stated tolerance, not treated as an aesthetic judgement.

The equipment decides whether the number is defensible

The measurement equipment is mundane and worth naming in the pack: a calibrated steel rule for lengths over 300 mm, a digital caliper for binding and piping width, a tape for the top oval only when a rule cannot reach, a 45 mm gauge for divider wells, a calibrated scale for weight, and a colour assessment under D65 with a physical standard. An inspection carried out with a cloth tape and a phone torch produces numbers that cannot be defended.

When a Lot Fails: Rework, Sort and Re-present

A failed lot is a process event, not a verdict on the supplier. The useful question is whether the defect is systemic or isolated, and that is answered by looking at where in the line the defective units came from. If the sample failures cluster in cartons packed on one shift, the cause is local. If they are spread evenly, the cause is in the specification or the materials and will recur on the next run.

Rework, sort or concession

The three remedies are different in cost and in evidence. Rework means the supplier repairs the identified units and the lot is re-presented for a fresh inspection, usually within 7-10 working days. A 100 per cent sort means every unit is inspected, defects are repaired or removed, and the lot is re-presented with a sorting record. A concession means the buyer accepts the lot at a discount, which is sometimes correct for a minor cosmetic issue on a promotional order and almost never correct for a functional one.

Re-presented lots should be inspected at the same level, on a fresh random sample, not on the units the supplier has set aside. The most common failure of the rework loop is an inspector shown the repaired units rather than a sample of the lot, which produces a clean report on a lot that has not changed.

On the money side, a failed inspection sits inside the 30/70 payment structure rather than outside it. The balance is not released until the lot passes, which is what gives the inspection its commercial force. A term sheet that pays the balance on a date rather than on a result removes the only leverage the report has.

The Limits of the Method

Sampling inspects a fraction of the lot and infers the rest. At AQL 2.5 on a 200-piece order, 32 bags are examined and 168 are not, so a lot can pass with a defective unit in an unsampled carton. That is not a flaw in the standard; it is the trade the standard makes, and it is the reason a buyer with a low tolerance for returns should invest in the process rather than in a lower AQL.

The three things that close the gap are all upstream. A pre-production meeting that walks the pack page by page, an inline check at the first 50 pieces where the line is still adjustable, and a final random audit from the buyer own side or a third party rather than the supplier quality team. Each costs less than a failed shipment and each addresses a cause that sampling can only detect.

Sampling also cannot see what only time reveals. A stitch that will fail after 200 insertions, a coating that will delaminate in a hot container, a zip that will corrode in a coastal climate: none of these show up in a one-day inspection. That is what the test protocol is for, run on the first article and on pre-production samples, with results recorded against the standards named in the pack.

For the wider quality system these inspections sit inside, the quality standards guide sets out how inline checks, final inspection and test protocols fit together, and how bag programmes keep quality consistent describes the process controls that make a passing lot the normal outcome rather than a lucky one.

Putting AQL Into the Purchase Order

An inspection standard that is not written into the commercial documents is a suggestion. Four sentences in the purchase order turn it into an obligation, and they take less time to draft than the argument they prevent.

First: the inspection basis. AQL 2.5 under ISO 2859-1, general inspection level II, pre-shipment, with the sample drawn randomly across the lot. Second: the defect list, attached as an annex with the classification of critical, major and minor and the tolerance for each measurement. Third: the consequence, stated as rework and re-presentation within a stated number of days at the supplier cost, with the balance withheld until a pass. Fourth: the remedy for a repeat failure, which is normally a 100 per cent sort at the supplier cost plus a right to cancel the balance of the order.

Two additions are worth the ink. Name who may inspect, so a buyer-appointed third party is not refused at the door. And state that the approved sample, the measurement chart and the defect list all travel with the inspection, because an inspector without the reference is measuring against memory.

None of this replaces a factory audit, and the two serve different purposes. An audit asks whether the process can produce a conforming bag; an inspection asks whether this shipment does. Buyers running both usually find that the audit predicts the inspection result, which is the cheapest possible early warning.

Sources and Standards Referenced

• ISO 2859-1:2022 Sampling procedures for inspection by attributes
• ISO 9001 quality management requirements for production control
• ANSI, American National Standards Institute, standards coordination
• ASTM D5034 Breaking strength and elongation of textile fabrics (grab test)
• ISO 105-B02 colour fastness to artificial light

Next steps: custom solutions, manufacturing capability, and a quotation request.

Frequently Asked Questions

What does AQL 2.5 mean in plain terms?

It is the acceptance quality limit used in a sampling plan. Lots running at or better than 2.5 per cent defective are accepted most of the time; lots substantially worse are rejected increasingly often. It is a rule for deciding what to do with a shipment, not a statement that 2.5 per cent of bags may be faulty.

How many bags are inspected in a 200-piece lot?

At general inspection level II with AQL 2.5, a 200-piece lot falls in the 151-280 band, giving code letter F, a sample of 32 bags, an acceptance number of two and a rejection number of three. Three defective units in the sample reject the lot.

How many bags are inspected in a 1,000-piece order?

A 1,000-piece lot falls in the 501-1,200 band, code letter H: 80 bags sampled, five accepted, six rejecting the lot. Note that the sample grows much more slowly than the lot, which is the efficiency of the scheme and also its blind spot.

What counts as a critical defect on a golf bag?

Anything affecting safety, legality or saleability: a needle fragment, mould or mildew, a sharp exposed moulding edge, a missing country-of-origin label or a required warning label. Critical defects are normally zero acceptance regardless of the AQL number.

What is the difference between a major and a minor defect?

A major defect affects function or is visibly wrong at retail: an open seam, a zip that will not run, a missing bar-tack, a colour mismatch outside dE 1.5, a logo in the wrong place. A minor defect is cosmetic and inside tolerance, such as a loose thread under 20 mm or topstitch uneven by less than 2 mm.

Can a lot pass inspection and still have problems?

Yes. On a 200-piece lot, 32 bags are examined and 168 are not, so a defective unit can sit in an unsampled carton. Sampling manages risk, it does not eliminate it. Process controls, an inline check at the first 50 pieces and a factory audit close the gap that sampling leaves.

Should the buyer or the supplier book and pay for the inspection?

The buyer should, whenever the report is going to release money. Letting the inspected side commission and pay for the visit creates the wrong incentive, and a man-day from an independent agency usually runs USD 300-500, which is small money against a five-figure shipment.

What happens if the lot fails?

The supplier reworks the identified units and re-presents the lot for a fresh inspection, usually within 7-10 working days, or runs a 100 per cent sort with a sorting record. The 70 per cent balance stays unpaid until a lot passes, which is what gives the report its commercial force.

Should the re-inspection use the same sample?

No. A re-presented lot should be sampled fresh and randomly across the cartons, not from the units the supplier has set aside for inspection. Showing the repaired units produces a clean report on a lot that has not actually changed.

Why is a 100 per cent sort sometimes better than rework?

When the defect is spread evenly through the lot rather than clustered in one shift, rework fixes what was found and leaves the rest. A full sort finds every instance and produces a record that can be checked, at a cost that is usually worth paying once on a lot with a systemic problem.

Does AQL 2.5 apply to the cartons as well as the bags?

The sample is drawn from cartons, so carton condition, labelling and unit count are inspected at the same time. Carton defects are usually classified separately from product defects, but a carton that cannot be scanned or that has crushed corners is a legitimate reason to hold a lot.

How do I write AQL into a purchase order?

State the standard and level, attach the defect list with critical, major and minor classes and measurement tolerances, state the consequence of failure including rework and re-presentation at supplier cost, and name who may inspect. Add that the approved sample and measurement chart travel with the inspection.