Piece-by-Piece (100%) Defect Sorting & Rework Supervision

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.

OUTCOME 01
Ship
Pieces that clear every check, counted, re-packed and sealed as a lot you can send without a second thought.
OUTCOME 02
Rework
Pieces with a fault that can be corrected without leaving a trace. Returned to the factory, supervised, re-inspected before they rejoin the lot.
OUTCOME 03
Reject
Pieces that cannot be brought to standard. Marked so they cannot quietly find their way back into the carton.
01 · The Number Vendors Leave Out

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.

80%
The share of defects a best-case human inspector catches under good conditions, per Sandia National Labs. Two inspectors working in tandem reach about 96% — not 100%. Other studies of repetitive visual inspection put the ordinary range lower, at 60 to 80%.

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.

INCOMING
5.0%
Defect rate in the batch before sorting begins, taken here as a worked example.
AFTER PASS 1
1.0%
One full pass at 80% detection. Already a large improvement, and still not shippable for most buyers.
AFTER PASS 2
0.2%
A second independent pass by a different person. Two in a thousand.
AFTER PASS 3
0.04%
A third pass, used on high-value lots. This is where “zero defect” becomes a fair description.
Illustrative arithmetic using the 80% figure above and a 5% starting rate, not a guaranteed outcome. Your real numbers depend on the defect types, the lighting, the pace set, and how long each inspector has been on the table. We quote the number of passes, not a promise of perfection.
02 · The Sort

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.

01

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.

WHAT IT PREVENTS
Two inspectors quietly applying two different rules to the same lot.
02

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.

WHAT IT PREVENTS
Borderline calls decided by recollection instead of by comparison.
03

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.

WHAT IT PREVENTS
Two sorters producing two different lots out of the same batch.
04

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.

WHAT IT PREVENTS
Detection sliding through the shift without anyone noticing it has.
05

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.

WHAT IT PREVENTS
The first inspector’s blind spots being repeated rather than covered.
06

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.

WHAT IT PREVENTS
Pieces moving between bins with no record that they moved.
07

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.

WHAT IT PREVENTS
A sorted carton and an unsorted one standing side by side, unlabelled.
03 · Three Bins, No Fourth

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.

SHIP

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
REWORK

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
REJECT

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
04 · Rework Supervision

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.

01
The repair method is agreed in writing first
Before a single piece is reworked, the method for each defect type is written down and approved — what is repaired, how, and what the result must look like. Without that, each operator invents a method, and a lot comes back repaired five different ways.
02
The first repairs are watched, not sampled
Our supervisor stands at the rework table for the first pieces of each defect type and approves the method in practice. Once the method is proven, the supervision moves to periodic checks, and the first approved piece stays on the table as the reference.
03
Repair, never conceal
Pressing a pucker flat, hiding a stain under a fold at pack, or re-bagging so the fault faces inward are not repairs. They survive the carton and arrive at your customer. Anything of that kind is recorded as a finding against the factory, not as a repaired piece.
04
Reworked pieces are re-inspected at 100%
Not sampled. These are pieces already known to have failed once, which makes them the least suitable population in the batch for statistical sampling. Each one goes back through the full check before it is allowed to rejoin the passed lot.
05
Rework stays physically separate until it passes
Repaired pieces are held in their own marked area, not returned to the general stock to be found later. The commonest way a defect reaches a carton is not a missed inspection — it is a reworked piece that was put back on the shelf before anyone re-checked it.
06
The cause goes back to the factory in writing
A sort that fixes the lot and teaches the factory nothing guarantees you will pay for another sort on the next order. The defect breakdown by type, operation and line goes back with the handover, so the conversation is about the cause rather than about the count.
05 · Handover

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.

DOCUMENT I

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.

DOCUMENT II

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.

DOCUMENT III

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.

DOCUMENT IV

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.

06 · Questions

Answered Before You Ask Them

Including the one about guarantees, which is the question this page exists to answer honestly.

01
Can you guarantee a zero-defect lot?
No, and nobody doing manual sorting can. Research on visual inspection puts a best-case human inspector at around 80% detection, with two working in tandem reaching roughly 96%. What we can commit to is the method: the number of independent passes, who performs them, the written standard applied, and a reconciliation that balances. A vendor who promises you 100% is either unaware of that evidence or hoping you are — and when a defect does reach your customer, the promise is what the argument will be about.
02
When is a 100% sort worth it instead of sampling?
Three situations. The batch has already failed an AQL inspection, so the sample has told you the lot is not acceptable and only piece-by-piece work can separate what is. The goods are valuable enough that a single escaped defect costs more than the sorting itself. Or the defect is one a buyer treats as critical, where the acceptance number is zero and sampling cannot give you that. Outside those three, sampling is usually the better use of the same money, and we will tell you so.
03
How fast is it, and how many people do you need?
It depends almost entirely on the garment and on how many checks each piece needs — a basic tee with three check points and a lined jacket with twenty are different jobs, and no honest per-piece rate covers both. We quote after seeing the style, the defect list and the quantity, and we quote the team size and the number of passes rather than a flat piece rate, because a piece rate quietly rewards speed over detection. Expect the realistic answer to be slower than the one you want.
04
Can this be done inside the factory?
Yes, and it usually should be, because that is where the rework happens and moving goods twice costs time you do not have before a ship date. What it requires is a dedicated area, our own lighting if the factory’s is uneven, and a written agreement that sorted and unsorted stock are kept apart. If the factory cannot give the space, sorting at a third-party facility works — but the rework loop becomes slower and should be priced that way.
05
Who decides what counts as a defect?
You do, or your buyer does, and it is written down before we start. We will propose a classification where none exists, based on the approved sample and the spec, but we do not want to be the party that invented the standard retrospectively. On a previously rejected batch this matters even more: the sorting rule should match the standard the lot was failed against, otherwise the sort passes pieces the buyer will reject a second time.
06
What stops a rejected piece from ending up back in a carton?
Physical separation and marking, done at the moment of rejection rather than at the end of the shift, plus a reconciliation that has to balance. It is the right question to ask, because the pressure to quietly return pieces to the lot is real when a ship date is close and the rejection count is high. The honest answer is that controls make it difficult and visible, not impossible — which is also why the reject count appears in your report rather than only in ours.
START HERE

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.