How to Read an AQL Inspection Report for Custom Bags

Reports arrive as a PDF somewhere between eight and forty pages, and the verdict on the cover — pass, fail, pending — is the least interesting thing in them. The verdict is a conclusion somebody reached about fifty or a hundred and twenty-five units. What you actually need is enough detail to agree with it, disagree with it, or learn something from it, and those three aims point at different pages.

What Does the Cover Page Tell You That You Need to Check First?

Start here, because an error in any of these five fields invalidates everything behind it. The product description and reference against your purchase order; the lot quantity presented; the quantity finished and packed at the time of the visit; the date and the location; and the inspection criteria as actually applied. Compare each to what you believe you ordered.

The one that most often differs is the finished quantity. Inspections are routinely run against a lot that is still being packed, which is fine and normal — provided the report says so. A plan executed against 320 finished units of a 500-piece order does not speak for the remaining 180, and the verdict should be understood as conditional. If the report is silent on the finished quantity, ask before releasing payment.

How Do You Verify the Sample Size Was Right?

Two numbers, side by side: what the plan required and what was drawn. Take the lot size from the cover, apply your inspection level, read the code letter, and check the sample size written in the report matches. Then check the number actually examined, which is sometimes lower because a carton could not be opened or a unit was short-shipped.

Discrepancies here are common and often innocent, but they matter because everything downstream scales with them. A plan that required 50 units and examined 32 does not fail at four defects; at code letter G the accept number is two. If the report does not reconcile the two figures, ask for an explanation in writing rather than accepting the verdict at face value.

Where Do the Defect Counts Sit and How Are They Read?

In the defect summary, which should list every instance separately with its class, a short description, a location, and a reference to a numbered photograph. Read three things out of it. The total per class against the accept number. Whether any instance is critical, which fails the lot at any count. And whether faults repeat — six instances of the same thing in the same place is a process finding even when the count passes.

Counts alone hide that last point, and it is the difference between a report you file and a report you act on. Two majors comprising one sticky zipper and one missing label is noise. Two majors comprising two zippers sticking in the same place, on units from the same carton, is a warning about the next lot. Ask for repetition to be flagged by category, not just counted.

A worked example shows how much a well-kept summary carries. The table below is a fifty-unit sample that passes on the major column with two counts but carries eleven minors, and it records three instances of one fault in three cartons. Read only the totals and you release the lot with a note about housekeeping. Read the locations and you have a finding about one station.

What Do the Photographs Prove and What Do They Hide?

Photographs are evidence of existence, not of class. They establish that something was wrong with a particular unit at a particular moment. They say nothing about how representative it was, and camera work can mislead in predictable directions: a close-up with no scale makes a 2 mm mark look like a gash, and studio lighting flattens colour differences that are obvious in daylight.

The discipline to insist on is a scale and a context shot for every individual close-up, together with the panel or component identified in frame. Where colour is the issue, ask for the unit beside the approved counter-sample rather than alone. A report whose photographs would not survive cross-examination by your own quality team is not worth paying for.

How Are Measurements Reported and Against What Limit?

Every number in the report should be accompanied by the limit it was read against and the method used to take it. "Length 46 cm" means nothing at all without those two companions. "Length 46.2 cm against 45.0 cm plus or minus 1.0 cm on the tech pack, measured flat on a bench with a steel tape to the outer edge of the binding" is a finding.

Watch particularly for measurements taken on a stuffed rather than an empty bag, dimensions measured to a seam rather than to a stated reference point, and fabric weights reported without the laboratory conditions they depend on. Each of those produces a number that is technically accurate and practically wrong, and each is avoided by naming the method in the tech pack rather than in the report.

What Does the Packing Section Say About What Actually Ships?

More than most buyers read. This is the section that tells you what went into the container: how many cartons, how many units per carton, the packing method and materials, the shipping marks and whether they match the ones on your documents, carton dimensions and gross weight, and any statement about the pallet configuration.

Discrepancies here are the ones that surface weeks later at a distribution centre, when a count is short or a carton weight does not match the manifest. A packing section that merely says "packed as per buyer requirement" is useless. Ask for the numbers, because a mismatched carton weight is often the first hard evidence that the shipment contents differ from what was inspected.

One field deserves particular attention where several styles ship together: mixed packing. Cartons holding more than one colourway or style are the usual source of destination-side miscounts, and they need to be declared on the report with the breakdown written inside each carton rather than averaged across the shipment. Ask for the mixed cartons to be listed individually.

Finally, read the weight. Gross weights per carton, the net figure and the number of units per carton together form an arithmetic check that takes thirty seconds and catches the two most common packing problems: a short count inside a sealed carton, and a substituted or mislabelled style shipped in the right box.

What Do You Do With Critical Defects Found in the Report?

Stop reading and act. Critical defects carry zero acceptance, so their presence fails the lot regardless of every other count, and the reason matters more than the verdict. A sharp burr on hardware and a substance over limit are both critical and they call for completely different responses: one is a hardware lot to be quarantined, the other is a question about the whole shipment and possibly the programme.

Ask for the units to be retained, and ask for the count rather than an instance. If one unit out of fifty carries a needle fragment, the question is not that unit but the control that let it through, and that question cannot be answered by looking harder at the remaining forty-nine. This is the one finding where the correct response is usually a process investigation, not a re-screen.

Why Does the Switching State Belong on the Cover?

Because it determines the plan used for the next lot and it is invisible otherwise. Normal, tightened, reduced or discontinued changes the accept numbers without changing the AQL you agreed, so a programme can be under tightened inspection and still report AQL 2.5 on every cover page. Without the state you cannot tell whether the numbers came from the column you specified.

Ask for two things: the state the programme was in at the start of this inspection, and whether this lot changed it. A vendor reporting back-to-back passes while sitting under tightened inspection is telling you something flattering that you would not otherwise know, and a run of passes under reduced inspection carries materially less evidence per lot.

How Should Reports Be Read Across Several Lots?

Keep them, sorted by date, with the switching state noted on each. A single report tells you about one lot; twelve of them tell you whether the programme is improving, which is information no single visit can provide. Plot three series: the total defect count per lot, the count in the two or three most repeated categories, and the ratio of minors to majors. The shape of those lines matters more than any individual verdict.

What you are looking for is direction and volatility. A programme whose counts drift upward over six lots has a maintenance or tool wear problem, not a quality one, and it will eventually produce a failed lot. One whose counts jump about has a control problem — inconsistent operators, unmanaged material changes, or a line that gets re-set every week. Both cost less to fix before the lot fails than after.

What Makes a Report Useful Six Weeks Later?

Two practical habits keep that record usable. File the photographs with the report rather than in an email thread three layers deep, and write one line at the top of each report recording what was agreed as a result — nothing was changed, a tolerance was clarified, a component was swapped. Those lines take thirty seconds and turn a stack of PDFs into a history somebody can actually read when a dispute arrives.

A report earns its cost when it can answer a question nobody thought to ask on inspection day: what did this lot look like, who decided it was acceptable, and on what basis. Three features deliver that. Every instance photographed with a scale and a context shot. Every measurement carrying its limit and its method. And a signature, with the inspector named.

Two references close the loop. Any outright disagreement about a count or a measurement is handled under the dispute route agreed before the visit, and the numbers themselves can be checked independently against the AQL sampling table. Reports asked for in this shape cost no more than the ones nobody reads — the difference is entirely in what is requested beforehand.

Frequently Asked Questions

People Also Ask