When full inspection beats sampling: part 2
Full inspection earns its fee when the cost of one escaped defect is higher than the cost of checking every unit. That is the short answer this second part builds on: sampling controls a stable process at a reasonable price, while piece-by-piece checking contains the damage after a process has already failed or when the product simply cannot be allowed to fail. Use this guide to recognise those moments, scope the sorting work sensibly, and combine it with the fixes that stop the same problem recurring in the next shipment.
Key takeawaysFull inspection is a containment action, not a quality strategy: it sorts good units from bad ones after something has gone wrong, while sampling monitors a process that is assumed to be under control. · Factory acceptance testing applies the same containment logic at system level, verifying that finished equipment meets the agreed criteria before it leaves the factory floor [1]. · The decision turns on defect cost: high unit value, safety relevance, or expensive returns push the maths toward checking every piece. · A documented quality management approach helps importers record what was found, what was sorted, and what corrective action the factory agreed, which is the purpose of the standards in the ISO 9000 family [2]. · Always pair a sorting job with a root-cause discussion, or the next shipment will need sorting too.
The first part of this series explained the basic trade-off: sampling inspects a fraction of a lot and accepts a small, quantified risk, while full inspection checks every unit and costs proportionally more. Most routine shipments sit comfortably on the sampling side of that line. This part is about the shipments that do not.
Those shipments usually announce themselves. A pre-shipment check fails badly, a customer returns a batch with the same fault, or a product with a safety function starts showing inconsistent behaviour. At that point the question is no longer whether the process is capable, because the evidence says it is not, and the buyer needs to know exactly which units are shippable.
What follows is a practical framework: where the cost crossover sits, which situations call for piece-by-piece checking, how to scope the work so it stays affordable, and how to make sure a sorting job is the last one rather than the first of many.
The cost crossover in plain terms
Sampling is priced per inspector-day and covers a defined sample size, so its cost is roughly flat regardless of lot size within a band. Full inspection scales with the number of units and the minutes each check takes, so it gets expensive fast on large lots of fiddly products. The crossover is found by comparing the sorting cost against the expected cost of letting defects through: return freight, replacement units, chargebacks, and the commercial damage of an unhappy customer.
A simple way to frame it is to multiply the observed defect rate by the lot size and then by the cost of one field failure. If that number is clearly above the quoted sorting cost, full inspection is the rational choice. If it is clearly below, sampling plus a corrective action plan is enough. The difficult cases sit in the middle, and those are decided by factors sampling maths cannot see, such as whether the defect is a scuffed carton or a faulty protective device.
Why defect type matters more than defect rate
A five percent rate of cosmetic marks on a metal enclosure is an annoyance that a discount or a rework bench can resolve. A one percent rate of a missing earthing connection is a different matter entirely. Sampling plans treat defects through classification systems with separate limits for critical, major, and minor issues, and the same thinking should drive the sorting decision: critical defects justify full inspection at rates that would never justify it for minor ones.
Stable process versus known failure
Sampling theory assumes the lot was produced by a stable process and uses the sample to confirm that assumption. Once a failed inspection or a customer complaint has broken that assumption, the statistical basis for sampling is gone for that lot. Sorting every unit rebuilds certainty directly instead of inferring it, which is why experienced quality managers stop sampling a lot the moment its process is known to have failed.
Situations where sorting every unit pays off
| Situation | Why sampling is not enough | What to sort for |
|---|---|---|
| A failed pre-shipment inspection with a high major-defect rate | The lot is known to be bad, so inference from a sample adds nothing | The specific failed checkpoints, unit by unit, plus a re-check of any repaired units |
| Safety-related or regulated functions | Even a small escape rate carries liability and recall exposure | Every safety checkpoint on every unit, with failed units quarantined and recorded |
| High-value goods with expensive returns | One return can cost more than sorting dozens of units | Function, appearance, and accessories against the approved specification |
| A new supplier with no track record on a complex product | There is no history to support the assumption of a stable process | Workmanship and functional checkpoints that reveal process control gaps |
| Customer complaints pointing to a systematic fault | The complaint defines the defect, so checking for exactly that fault is efficient | The reported fault across the whole suspect lot, then an agreed rework standard |
Notice what these rows have in common: in each case something has already told the buyer that the lot cannot be trusted. Full inspection is the response to evidence, not a default posture, and that distinction keeps inspection budgets under control.
Scoping a sorting job so it stays affordable
- Define the defect list narrowlySort for the known faults and the checkpoints around them, not for the entire specification. A tight defect list can cut sorting time per unit dramatically.
- Agree the disposition of each categoryDecide in advance what happens to good units, reworkable units, and scrap: who reworks, who re-checks, and who pays. Write it down before the sorters arrive.
- Prepare the floor space and the toolsSorting needs tables, power for functional tests, calibrated gauges where relevant, and marked zones for pass, rework, and reject. A system-level check at the end, in the spirit of factory acceptance testing, confirms the sorted lot as a whole [1].
- Record results lot by lotKeep counts of checked, passed, reworked, and rejected units with the serial or carton references. This record is what makes the rework billable and the corrective action verifiable.
- Book the follow-up before the sorters leaveSchedule the root-cause meeting and the next inspection while everyone is still on site. Memory fades fast once the shipment has sailed.
Running the sort without stopping the factory
A sorting job lives or dies on cooperation with the factory. The clearly stated position is that sorting protects both sides: the buyer avoids shipping defects, and the factory avoids a rejection of the whole lot. When neither the buyer nor the factory has spare hands for a multi-day sort, many importers ask a provider such as UTS Inspection to run the sorting inspection on site and report results lot by lot.
Keep sorted and unsorted stock physically apart from the first hour, because mixing them silently invalidates the whole exercise. Label the zones, control who moves cartons between them, and have the rework bench feed its output back through the checking station rather than straight into the pass pile. These sound like details, but most disputed sorting jobs fail on exactly these points rather than on the checking itself.
Document the quality loop
Quality management standards in the ISO 9000 family emphasise documented procedures, records, and corrective action rather than any single checking technique [2]. A sorting report that lists what was found, what was done with each category, and what the factory committed to change fits that model directly, and it gives the buyer leverage if the next lot shows the same fault.
A useful rule of thumb: if you cannot name the specific defect you are sorting for, you are probably not ready for full inspection. Go back one step, run a focused inspection to characterise the problem, and then scope the sort around what it found. Sorting without a defect list is the most expensive way to learn what sampling would have told you for a fraction of the price.
Sources
- Wikipedia: Factory acceptance test —— Overview of factory acceptance testing performed before shipment or installation.
- Wikipedia: ISO 9000 —— Overview of the ISO 9000 family of quality management standards.
Frequently asked questions
How much more does full inspection cost than sampling?
It depends on lot size and check duration per unit, but a full sort typically costs several times a sampling inspection on the same lot because every unit consumes inspector minutes. The right comparison is not against the sampling price but against the cost of the defects that would otherwise ship: returns, replacements, and customer damage. Get a quote for both and run the crossover calculation described above.
Can a sorted lot still fail a second inspection?
Yes, if the sort was poorly controlled: mixed stock, unchecked rework, or an incomplete defect list all let defects through. A well-run sort with segregated zones and re-checking of reworked units should pass a follow-up sampling check comfortably. That follow-up check is worth booking precisely because it verifies the sort worked.
Who pays for sorting when the factory caused the defects?
Commercial terms decide, not inspection standards, so the purchase contract and any quality agreement matter here. Many buyers negotiate that the supplier bears sorting and rework costs when defect rates exceed an agreed threshold. Whatever the arrangement, agree it in writing before the sorters arrive, because after-the-fact billing for sorting is a frequent source of disputes.
Does full inspection replace the need for process improvement?
No, and treating it as a substitute is the classic trap. Sorting removes the bad units from one lot but changes nothing about how the next lot will be made. The sorting report should feed directly into corrective action: which process step produced the fault, what the factory will change, and how the next inspection will verify the change.
The messy middle is where the actual life is happening — the part no one puts on a vision board. — From the Perfectly Cursed Life editorial line
Join the Cursed Club.
One candid essay in your inbox every Sunday morning. Reader-funded, ad-free, and written for humans still figuring it out.
Join the Cursed Club — Free Newsletter