Almost every first order is inspected once, at the end, because that is what "inspection" means to most buyers. It is a reasonable default and it is often the wrong single choice — not because the pre-shipment inspection is weak, but because by the time it happens nothing about the product can be changed cheaply any more.
The only question worth asking
At each point in the production timeline, ask: what can still be changed, and what does changing it cost? Everything else follows from the answer.
- Before production starts. The inputs are still changeable. Wrong fabric weight, wrong steel grade, wrong component — fixing it is a purchase order amendment and a delay measured in days.
- Somewhere between 20% and 80% complete. The method is still changeable. The line is set up, the operators are in rhythm, and a construction fault can be corrected for the remaining 80% while the finished 20% gets reworked.
- Finished and packed. Nothing about the product is changeable. Your options are sort, rework everything, discount, or reject — all of which mean handling every carton again.
- At the container door. Nothing at all. You are verifying, not fixing.
Which produces the trade-off nobody states out loud: the earlier an inspection is, the more it can change and the less it can tell you. A pre-production check cannot report a defect rate because there is nothing made yet. A pre-shipment inspection reports the defect rate precisely, at the moment you can do least about it.
What each one uniquely sees
The useful comparison is not what each inspection covers — they overlap heavily — but what each is the only one able to tell you.
| Stage | Only this one can | Cannot tell you |
|---|---|---|
| Pre-production | Confirm the materials and components going in, while substituting them is still cheap | Anything about workmanship — nothing is built yet |
| During production (20–80%) | Catch a construction fault while the line is still running and the fix applies to the rest of the order | Final quantity, packing, or labelling — none of it exists yet |
| Pre-shipment | Verify quantity, packing, labelling and the finished defect rate on goods as they will ship | What was used to build it, and what happens after the inspector leaves |
| Container loading | Confirm what physically goes into the container, in what condition and what order | Anything about product quality that was not already established |
Which to buy, by situation
Inspection budget is finite. These are the pairings that make sense in practice.
- First order with a new factory: during production plus pre-shipment. Not pre-shipment alone. You have no history, and the during-production visit is the only one that can still change the outcome of this order rather than just report on it.
- Repeat order, same factory, unchanged product: pre-shipment alone. The history is your early-warning system. Spending on a during-production check every time is the most common way to overspend on quality control.
- New product at a factory you already trust: during production matters more than pre-shipment. The relationship is proven; the construction is not.
- Anything with a certification or safety dimension: add pre-production. Material substitution is invisible in a finished product and expensive to discover at a lab after shipping.
- High value, consolidated loads, or several SKUs in one container: add loading supervision. It is also the stage at which the verified gross mass of the packed container is established under the SOLAS requirements — a useful moment to have someone of yours present.
- Schedule with no slack: during production becomes more important, not less. This is counterintuitive and it is the one buyers get wrong most often. If the pre-shipment inspection fails and there is no time to rework, the inspection has cost you money and bought you nothing but a difficult decision.
The hidden cost is the calendar, not the fee
An inspection is roughly a day of an inspector's time. The expensive part is the slack it requires around it.
A pre-shipment inspection needs to happen with enough time before the terminal cut-off for a failure to be survivable — which means production genuinely complete, then the visit, then a realistic rework window, then the goods still making the booking. Buyers who plan the inspection backwards from the ship date rather than forwards from the production-complete date routinely end up inspecting a half-finished order or accepting a marginal result because the alternative is missing the vessel.
That planning is the same arithmetic as the rest of the shipping timeline, and it is worked through in what sea freight transit times are actually made of.
Before booking any of them: verify the stage
The most common wasted inspection is not the wrong type. It is the right type, booked against a progress report that was optimistic.
Factories report progress to the person chasing them, and "nearly finished" is a relationship statement as much as a production one. An inspector arriving to find 15% complete cannot do a during-production check, and one arriving to find goods unpacked cannot do a pre-shipment inspection. Either way the visit is paid for and has to be repeated.
Two cheap habits remove most of it:
- Ask for dated evidence, not a percentage. Photographs of the line with output visible, or of stacked finished cartons, taken the day you ask. A factory genuinely at 40% can produce that in ten minutes; one that is at 15% will offer words instead.
- For a pre-shipment inspection, ask for the packing list first. If the packing list does not exist, the goods are not packed, and a pre-shipment inspection cannot verify quantity or carton data — which is half of what you are paying it to do.
This is also why the inspection date should be tied to production completion rather than to the vessel. A booking made against the ship date silently assumes the factory will be ready, and that assumption is the one that fails most often.
What it costs against what it prevents
An inspection is about a day of one person's time. The comparison that matters is not that fee against your margin, but that fee against the cost of the failure it would have caught.
Take the example from the sampling guide: a lot that passes at AQL 2.5 having shown 8 major defects in a sample of 200. If the sample is representative, roughly 4% of a 5,000-piece order has a major defect — about 200 units. What happens to those 200 units is where the money is:
- Caught during production. The factory reworks them and corrects the remaining run. You lose days, not units.
- Caught at pre-shipment. Sorting 5,000 pieces to find 200 means handling every carton in the order, and somebody is paying for that labour and the delay it causes.
- Caught at your warehouse. Now the 200 units have crossed an ocean. You have paid freight and duty on defective goods, and sorting happens at destination labour rates instead of factory ones.
- Caught by your customer. Returns, refunds, reviews, and on some marketplaces a defect rate that threatens the listing itself.
Each step to the right multiplies the cost of the same 200 units. That escalation, not the inspection fee, is what you are buying against — and it is why the cheapest inspection is rarely the one that saves the most money.
Who should do the inspecting
Three options, with real trade-offs rather than one obvious answer.
- The factory's own QC. Not worthless — they catch things early and they know the product better than anyone. But they are reporting on their own output to the person paying them, and no amount of good faith removes that structure. Treat their reports as useful information, not as verification.
- An independent third-party inspection company. Structurally independent, works to the standard, produces a report that will hold up in a dispute. The limitation is context: they do not know which of your requirements are commercially critical and which are nice to have, so they apply the checklist you gave them and nothing more. Which brings you back to the quality of that checklist.
- Your sourcing agent. Knows your product, your history and what actually matters commercially, and can act on findings immediately rather than emailing a report. The conflict is real and worth naming: an agent who both selected the supplier and inspects the output has an incentive not to fail their own choice. The honest test is whether failed inspections ever actually happen, and whether you receive the report unedited regardless of what it says.
For a first order with a new factory, a third-party report plus your own reading of it is the conservative choice. For an established programme, the speed of someone who can walk the line and resolve a problem the same afternoon usually outweighs the theoretical independence — provided you are getting the reports raw.
What none of them replace
All four are sampling exercises governed by the same standard, so all four inherit its limits: a pass is the absence of a reason to reject, not a guarantee of quality, and the accept numbers come from ISO 2859-1 rather than from anyone's judgement about your product. The mechanics are in AQL sampling explained, and what to specify so that any of these visits is worth paying for is in the inspection checklist.
More inspections also do not fix a supplier who cannot hold the specification. If a factory needs a during-production visit, a pre-shipment inspection and a re-inspection on every order to reach acceptable output, you are paying for quality control that the factory should be doing, and the honest conclusion is a supplier conclusion.
