How Many Units Get Inspected in a 500-Piece Bag Order?

The question arrives in almost every first enquiry, usually phrased as though there were a rate involved. There is not. Nothing in ISO 2859-1 asks what proportion of the order should be opened. The plan takes the quantity presented for inspection, converts it to a code letter through a table, takes the sample size written against that letter, and then reads the defects found in that sample against an accept number set by the AQL.

Why Is It 50 Units and Not 500?

Because inspecting everything is expensive, slow, and for most soft goods unnecessary. A full check on 500 sewn units costs a large multiple of a samples visit and still depends entirely on what somebody decides counts as a fault. Acceptance sampling accepts a measured, bounded risk instead: you give up certainty about the units you did not open, in exchange for a verdict that costs one inspector-day and can be applied to every lot the same way.

The economics are not linear, and that is deliberate. Notice what happens as orders grow: a 500-piece lot draws 50 units, a 3,000-piece lot draws 125, and a 30,000-piece lot draws 315. Ten times the order does not mean ten times the inspection. The plan is buying evidence about how the process is running, not coverage of every carton, and the statistical return on additional units falls away quickly.

It is also worth being clear about what those 50 units protect against, because expectations tend to run ahead of the method. A sample inspection detects faults that are spread through a lot: a machine set wrong, a fabric roll with a consistent defect, a printing screen that drifted. It detects clustered faults poorly, and defects that appear after packing not at all. Nothing in a sampling plan protects against a carton dropped on the way to the container, and nothing in it tells you anything about a unit that left on last month's shipment.

How Are Those 50 Units Chosen?

At random, from across the whole lot, by the inspector and not by the production floor. This single rule carries more weight than many buyers realise. A random sample only supports a verdict about the lot if every unit in the lot had a genuine chance of being drawn. The moment selection is helped — the nearest cartons, the top layer, the units nearest the door — the sample stops representing anything.

The practical instruction is to select from a carton list rather than from a pile. Ask for the packing list first, choose the carton numbers, have those cartons brought to the inspection area, then choose within them. If the floor cannot produce a carton list, that is itself a finding worth recording: it means the lot cannot be sampled properly and the verdict cannot be trusted.

The same logic applies to replacements. Where a unit is withdrawn because it failed during destructive testing, the replacement should come from a different carton than the one drawn from earlier, otherwise an unlucky carton supplies several chances to fail. It is a small rule, and it removes a genuine grievance when a second visit becomes necessary.

What Happens to the Units That Are Opened?

They were yours before the inspection and they are yours after it, but their condition changed. Polybags get cut, care labels get unfolded, tape gets removed, and some tests are destructive. The instruction should say who repacks, what materials are used, and whether the repacked units ship as first-quality goods or are replaced. Most arrangements replace them, which costs little at 50 units and removes an argument about repacking quality.

Where destructive testing is part of the plan — a strap pull, a weld burst, a coating adhesion test — the destroyed units are simply gone, and the cost of those units belongs to whichever party the contract assigns it to. Small numbers at 50 units, but the principle should be written down once rather than negotiated per visit.

How Long Does Inspecting 50 Units Take?

Most of a working day for one inspector on a general-purpose bag with a written checklist. The measurement work — dimensions, GSM checks, colour comparison, hardware function, stitch density — dominates the elapsed time, not the visual work. Expect the report to follow within one working day of the visit, longer where physical tests are sent to a laboratory.

Book the visit against packing progress rather than against a date. The inspection can only run when the lot is finished and packed; asking for it before the last cartons exist means either a wasted visit or an inspection of part of the order that does not represent the whole. The usual trigger is 80% packed with the remainder expected within the week.

Two practical arrangements make the day go smoothly, and both are worth settling in the instruction rather than on the phone. The first is who attends: an inspection attended by nobody from either side tends to produce a report nobody can question, which sounds efficient until something is wrong. The second is access — a table in daylight, a power supply, a scale, and someone who can open cartons. Work done on a bench in a corridor misses the faults that only appear under good light.

How Does the Count Translate Into a Pass or Fail?

Each defect found is counted into a class, and each class is read against its own column. On a 50-unit sample: a single critical defect fails the lot outright; three or fewer major defects pass it and four fail it; ten or fewer minor defects pass it and eleven fail. Three independent verdicts on one sample, and any one of them failing fails the lot.

Note what is being counted. The default is defects rather than defective units, so one unit carrying a printed logo error and a skipped stitch contributes two to the major count. If you would rather count affected units, say so in the specification — it is a legitimate convention, but it loosens the plan and applying it retroactively after a failure is not a serious argument.

Does Splitting the Order Into Smaller Lots Reduce the Sample?

No, and this surprises people. Inspection is applied per lot presented, so splitting 500 pieces into five shipments of 100 asks for five separate inspections at code letter F, which is 20 units each — a hundred units examined in total instead of fifty. Split into two lots of 250, the arithmetic gives code letter G twice, 32 units each, and sixty-four in total. Dividing a lot makes the inspection larger, not smaller.

The reason it feels as though it should help is that each individual sample is smaller, and therefore more likely to contain no defects. That is exactly the point: each lot is easier to pass, and five easy verdicts replace one harder one. Where the shipping schedule genuinely requires splits, that is commercial reality. Where a split appears just before inspection and after the errors were found, it is a tactic, and the purchase order should require written approval for it.

What Do 50 Units Statistically Detect?

Broadly: a perfect sample is strong evidence the process ran well, and a bad sample is conclusive evidence it did not. What the plan cannot do is distinguish fine gradations. A plan of 50 units with accept 3 takes a lot at about 2.5% most of the time, takes considerably worse lots sometimes, and still passes a poor lot occasionally. Every acceptance sampling plan has this shape; the curve can be tightened by examining more units, never by arguing about the accept number.

The practical reading for a buyer is that a passed lot is a probabilistic statement, not a warranty. If the consequence of a defective unit reaching a customer is severe — a safety exposure, a regulatory filing, a retail chargeback that exceeds the order value — then sampling is the wrong tool and a full check, a laboratory test or a different process control belongs in the plan instead.

There is one refinement worth knowing about, because it improves detection at no extra cost. The plan assumes defects are spread evenly, so its real-world performance is governed by how true that assumption is. Drawing units from several cartons rather than from one improves it at no extra cost; drawing 32 units from one carton and 18 from its neighbour does not. Ask for the sample to be spread, and you get closer to the curve the standard describes without paying for a larger sample.

What Should the Record Show Besides the Count?

Rather more than the count, if the report is any use later. A well-built report records the carton numbers sampled, the defect photographs with locations, the measurements taken and against which specification limit, the environment the inspection ran in, and the number of units examined against the number planned. Those details turn a verdict into evidence that survives a disagreement six weeks later.

Two fields are frequently missing and should be asked for on every visit. The first is the quantity actually finished and packed at the time of inspection, which may be less than the lot — a plan executed against part of the order does not speak for the whole. The second is whether the switching state changed as a result of this lot, since that determines what happens on the next one.

How Does This Fit the Programme Calendar?

At the end of the bulk window, never inside it. Sampling runs 6-10 working days from a signed-off tech pack, bulk production runs 35-50 days from written sample approval, and the inspection belongs after those, once the goods are finished and packed. Building the inspection into the production days assumes a clean result on the first attempt, which is the assumption that puts shipment dates at risk.

The prudent plan adds a contingency for the failure case. A first failure costs a re-work cycle plus a complete second inspection, and because the second draws a fresh sample it carries all the statistical characteristics of the first. Programmes with a fixed retail date should agree now what happens then: re-work and resubmission, a sort and partial release, or a signed waiver, and who pays for each.

Two working references help before any of this reaches a purchase order. The sample size for any other quantity can be checked with the AQL sampling table, and the reason inspection sits at the end of the window rather than inside it is set out in why the bulk window runs 35-50 days. Both are worth reading once, before a date is promised to anybody.

Frequently Asked Questions

People Also Ask