Every piece checked. Not every defect found.
A 100% sort is what you order when a batch has already failed, or when the goods are too valuable to gamble on a sample. It is the right call. But one manual pass over every unit catches roughly four defects in five — so what decides the outcome is not that everything was checked, it is how many times, and what happens to the pieces that fail.
Why One Pass Is Never The Whole Answer
Most sorting quotes promise a zero-defect lot after a single 100% check. The research on human inspection does not support that promise, and it is better to know it before the goods are on a vessel.
The limit is not skill or effort. Visual inspection is repetitive, and attention falls away within the first half hour of it — which is why an inspector who is sharp on carton one is measurably less sharp on carton forty. No amount of instruction removes that.
Once you accept the number, the design of the job changes. A single 100% pass is not a wall; it is a filter with a known leak rate. The way to close a leak like that is not to inspect harder, it is to inspect again, independently, and to put the pieces that fail through a loop that ends in another inspection rather than in the carton.
That is the whole reason this service is built around two passes and a supervised rework loop instead of one heroic sweep. The arithmetic below is why.
Seven Steps, In This Order
The order matters more than the speed. Most sorting that goes wrong goes wrong in the first two steps, before a single garment has been picked up.
Agree the standard before anyone starts
A sort is only as good as the line between pass and fail, and that line has to be written down before it is applied to twenty thousand pieces. We take the approved sample, the spec sheet and the buyer’s defect classification, and produce a one-page accept and reject rule for this lot.
Build the sorting station
Consistent lighting, a clear table, and a physical sample board showing an accepted piece next to a rejected one for each defect type. The reason is unglamorous: inspectors disagree far more when they are comparing a garment to a memory than to an object in front of them.
Calibrate the team
Before the real sort, the team inspects a small set of pieces seeded with known defects and known good units. Where calls differ, the rule is clarified on the spot. This is the cheapest half hour in the whole job.
First pass — every unit
Each piece handled individually against the written rule, with defects recorded by type and location rather than counted as a single total. Pace is set deliberately and breaks are scheduled, because an unbroken shift is how detection rates fall.
Second pass — different person
Everything that passed the first pass goes through a second, by someone who did not do the first. Independence is the point; a re-check by the same inspector repeats the same blind spots and adds very little. On high-value lots a third pass is added.
Count and reconcile
Pieces in, pieces passed, pieces reworked, pieces rejected — reconciled so the four numbers agree. A sort where the arithmetic does not close is a sort where something has moved between bins unrecorded, and that is worth stopping for.
Seal, mark and hand over
Passed goods re-packed, cartons marked so a sorted lot cannot be confused with an unsorted one, and rejects physically separated and marked. The handover report states the counts, the defect breakdown and how many passes were run.
Where Each Piece Ends Up
This service inspects every unit of high-value or previously rejected batches to segregate zero-defect goods for shipment and supervise factory rework. Every piece therefore leaves the table in one of three states — and there is deliberately no “probably fine” bin, because that is the bin defects travel in.
Clears every check
Meets the written standard on construction, measurement, appearance, labelling and packing. Counted, re-packed and sealed as a sorted lot, with the carton marked so it cannot be mixed back in with unsorted stock.
- Re-counted at pack, not estimated
- Cartons marked as sorted and dated
- Photographed as packed
- Count stated in the handover report
Correctable without a trace
A fault the factory can repair so that the finished piece is indistinguishable from one that never failed. If the repair would leave a visible mark, a pucker or a shadow, the piece is not a rework candidate — it is a reject.
- Re-stitching an open or skipped seam
- Replacing a button, snap or label
- Thread trimming and re-pressing
- Spot cleaning within an agreed limit
Cannot be brought to standard
Fabric faults, holes, shade mismatch, measurement outside tolerance, or anything safety related. Marked so the piece cannot re-enter the lot, held separately, and counted in the report rather than quietly absorbed.
- Marked at the point of rejection
- Stored apart from passed goods
- Counted and photographed by defect
- Disposal agreed with you, not assumed
The Half Of The Job That Is Not Inspection
Sorting identifies the problem. Rework is where it is either solved or disguised, and the difference between those two is almost entirely a question of who is standing there while it happens.
What You Get When The Table Is Cleared
Four documents, written so you can forward them to a buyer who is already unhappy about the first inspection.
Sorting certificate
Lot, quantity, dates, how many passes were run, who ran them, and the written standard applied — the single page a buyer asks for first.
Defect breakdown
Every defect found, grouped by type, operation and location, with counts and photographs. The ranking is what tells you where the fault actually started.
Reconciliation sheet
Pieces in, passed, reworked, re-inspected and rejected, balanced so the arithmetic closes. If it does not close, we say so rather than adjusting a figure.
Rework record
Each repair method as approved, which pieces went through it, and the re-inspection result for every one of them — the evidence that the loop was actually closed.
Answered Before You Ask Them
Including the one about guarantees, which is the question this page exists to answer honestly.
01Can you guarantee a zero-defect lot?
02When is a 100% sort worth it instead of sampling?
03How fast is it, and how many people do you need?
04Can this be done inside the factory?
05Who decides what counts as a defect?
06What stops a rejected piece from ending up back in a carton?
Send the failed inspection report and the piece count.
Those two decide almost everything — how many passes the lot needs, how many people, and whether sorting is even the right answer or whether the batch should go back to the line. If it is the second, we would rather say so now than invoice you for a sort that will not fix it.
