Spot Cleaning and Rework: The Hidden Cost Between Sewing and Packing

Ask a production manager how much rework the factory did last month and you will usually get a shrug and a guess. Ask how many pieces were sewn and you will get a number to the unit. That asymmetry is the whole problem. Rework is real work, paid for out of the same wage bill, occupying the same machines, and in most factories it is the only substantial category of work that nobody counts.

I run a CMT factory in Nepal. We counted sewing output precisely for years while the repair basket under the checking table was managed entirely by memory. This article is about what we found when we started counting it, and the mechanism that made counting possible.

Where the Defects Actually Come From

Rework in the finishing area is a mixture of things that arrived from different places, and treating it as one category is why it never gets fixed. Broadly it splits four ways.

Fabric-borne defects arrived inside the roll: slubs, weaving faults, a colour shade variation that only shows when two panels sit next to each other. No operator caused these and no training will prevent them. They are an inbound inspection problem, which is why the 4-point fabric inspection system pays for itself even though it feels like a delay at goods-in.

Handling stains are the most common category in my experience and the most preventable. Machine oil from an over-oiled machine head, grease from a trolley wheel, dirt from a garment that touched the floor, marks from a hand that just handled a chalk pattern. Every one of these is a housekeeping fact rather than a skill fact, and they cluster around specific machines and specific tables in a way that becomes obvious the moment you record which lot and which line they came from.

Stitching defects are the classic operator-and-machine category: skipped stitches, broken stitches, open seams, puckering, uneven topstitching. These attribute cleanly, because the bundle scan records the operator, machine and operation. Whether the answer is training or a mechanic depends on whether the defects follow the operator or follow the machine, and that is a question the scan data answers directly.

Measurement and construction faults are the expensive ones: a panel cut slightly off, a wrong label sewn in, sleeves attached with an asymmetry that only shows at the checking table. These often trace back past sewing into cutting or pattern, and they tend to arrive in quantity rather than singly, because whatever caused them caused it for a whole layer or a whole bundle.

Why Rework Goes Invisible

A defective piece leaves the tracked world the moment it is pulled out of the flow. Consider what happens in a factory without a rework mechanism:

• The checker finds a skipped seam and puts the garment in a basket.
• The piece has already been recorded as sewn, so production reports count it as output.
• It has not been packed, so dispatch does not know it exists.
• The operator who will repair it is not assigned to it in any system, so nobody is accountable for the basket.
• Two weeks later the lot closes short by the size of the basket, and someone empties it in a panic.

This is invisible work in progress, and it is worse than ordinary WIP because ordinary WIP at least sits in a bundle with a QR code on it. The repair basket has no identity at all. The general problem of WIP you cannot see is covered in our guide to WIP tracking; rework is its most acute form.

The Repair Bundle Mechanism

The fix we settled on is simple and slightly boring, which is usually a good sign. Damaged pieces are split out of the original bundle into a repair bundle, and that repair bundle gets its own QR code.

That one decision does several things at once. The repair bundle appears in the work pool like any other work, so it can be assigned to an operator rather than waiting for someone to notice the basket. It can be scanned at start and finish, so the time it took is recorded. It has a quantity, so the queue has a number rather than a volume. And when it completes, the pieces rejoin the flow through the normal path instead of being carried across the room by hand.

The damage report that creates the repair bundle carries photo evidence and a severity level: minor, moderate, major or critical. The photo matters more than I expected when we started. It settles, in about four seconds, the argument about whether the hole was in the fabric before the operation or made during it. Without a photo, that argument happens weeks later at the pay window with no evidence on either side and both people convinced they are right.

Severity also drives the payment-hold rules. On major and critical damage, payment for the work involved is held while the report is reviewed; on moderate damage a portion is held. The point of putting the rule inside the software rather than leaving it to a supervisor's discretion is consistency. Operators will accept a penalty rule that applies identically to everyone and will not accept one that appears to depend on who is asking.

Why Repair Pays 1.5×

The repair rate is the part of this that surprises people, so it is worth explaining properly. In our system repair work pays 1.5 times the operation rate.

Under piece rate, an operator choosing between a fresh bundle and a repair bundle is choosing between predictable earnings and an unknown. A fresh bundle of thirty pieces is thirty identical operations at a known pace. A repair bundle of thirty pieces is thirty different problems: unpicking somebody else's stitching, matching thread, handling a garment that is already pressed and must not be marked. If both pay the same per piece, the rational operator takes the fresh bundle every time, and the repair queue grows regardless of how many times you ask people to clear it.

The multiplier is not generosity. It is the price of getting the queue cleared today rather than at lot close. A repair that sits for three weeks eventually costs an air freight surcharge or a discount on a short shipment, and both of those dwarf the wage difference. If you want to see the arithmetic of what a short shipment costs, it is in the dispatch accuracy article.

Repair or Downgrade: Making the Decision Quickly

The second half of the discipline is knowing when not to repair. Some pieces should go straight to B-grade, and the cost of dithering is higher than the cost of being slightly wrong.

The general practice in most CMT factories is to weigh three things: whether the defect is repairable at all, how much operation time the repair consumes, and whether the repaired piece will pass the buyer's final inspection. A skipped seam on a side panel is worth repairing. A needle hole in the middle of the front panel of a plain jersey tee is not, because there is no repair that survives inspection. A stain that has been through the press has usually set and is a poorer candidate than the same stain caught before pressing, which is one practical reason spot cleaning should sit before the iron rather than after it.

We track B-grade inventory by defect type: stain, tear or hole, size issue, stitching defect. That categorisation is not administrative tidiness. The four categories have genuinely different destinations. A stain may recover on a second cleaning attempt. A tear or hole will not recover at all. A size issue is a fully well-made garment that happens to be the wrong size and can often sell as a second at full construction quality. A stitching defect may be repairable later when the line is quiet.

The Real Cost of Rework

The reason rework deserves more attention than its piece count suggests is the way its cost compounds. Follow one garment that needs a seam redone.

The standard minutes to make it were spent once, in the normal flow. The piece then consumes unpicking time, re-sewing time at the repair rate, re-checking time, and usually re-pressing time because the garment was already pressed when the defect was found. So the operation is paid for twice while only one saleable garment emerges. That is the core of it: the minutes are consumed twice and the output is counted once.

On top of that sit costs that do not appear on any operation card. The machine time the repair occupies is machine time not spent on new work, which matters most when the line is at capacity, which is exactly when rework peaks. The supervisor time spent finding the piece, deciding its fate and chasing the operator is unbudgeted. And if the repair queue is old enough that the lot ships short, there is a commercial cost that has nothing to do with the wage bill at all.

I am deliberately not quoting a multiplier here. I have seen figures in the trade press ranging widely and I have no basis for endorsing any of them. What I can say is the structural fact: minutes twice, output once, plus unmeasured supervisor and machine time. Once you record repair bundles, your own factory produces your own number, which is the only one worth planning against.

Making Rework Visible Without Adding Paperwork

What we settled on: the checker scans the bundle QR, which already carries the lot, article, colour, size, component and quantity, records the defect type and severity, and takes a photo if there is damage to attribute. The repair bundle and its QR come out of that. Nobody writes anything down and nobody types a tally into a computer in the evening. The rest of the design principles behind that are in the piece on garment QC software, and the numbers it produces feed the quality dashboard and your DHU, which has its own benchmark scale.

Reducing Rework: The Unglamorous List

• Inbound fabric inspection. The cheapest defect to deal with is the one you refuse before cutting.
• Machine oiling discipline and trolley maintenance. Oil stains and wheel grease are housekeeping, and they cluster in identifiable places.
• Light at the checking table. Bad light produces both missed defects and false rejections, and the second is as expensive as the first.
• Spot cleaning before pressing, not after. Heat sets many stains. The sequence matters more than the chemistry.
• In-line checking, not just end-line. A defect caught at operation four costs four operations of rework. Caught at the finishing table it costs the whole garment.
• Clear the queue daily. The number to publish in the morning meeting is the age of the oldest repair, not the total.

Frequently Asked Questions

What is rework in garment manufacturing?

Rework is any work done a second time on a garment that has already been sewn: unpicking and re-sewing a seam, replacing a wrong label, spot cleaning a stain, re-pressing a piece that was handled after pressing. It is distinct from scrap, where the piece is written off, and from B-grade, where the piece is accepted at a lower grade rather than fixed. In most factories it is the largest category of work that nobody counts.

Why does rework cost so much more than its piece count suggests?

Because the standard minutes are consumed twice and only one saleable garment emerges. The piece was already made once in the normal flow; the repair then consumes unpicking time, re-sewing time, re-checking time and usually re-pressing time, while output is still counted once. On top of that sit machine time not spent on new work, supervisor time spent finding and chasing the piece, and the commercial cost if an aged repair queue makes the lot ship short. The structural point is minutes twice, output once — the actual multiplier is specific to your factory and worth measuring rather than assuming.

How does a repair bundle work?

Damaged pieces are split out of the original bundle into a repair bundle that gets its own QR code. Because it has a code, it appears in the work pool like any other work, can be assigned to a named operator, can be scanned at start and finish, and has a countable quantity instead of being a basket under a table. The damage report that creates it carries photo evidence and a severity level of minor, moderate, major or critical, and severity drives the payment-hold rules.

Why pay 1.5 times the rate for repair work?

Under piece rate, an operator offered a fresh bundle and a repair bundle at the same rate will take the fresh bundle every time, because a fresh bundle is thirty identical operations at a known pace and a repair bundle is thirty different problems on a garment that has already been pressed. Without a premium the repair queue grows no matter how often it is mentioned in the morning meeting. The 1.5 multiplier is the price of clearing the queue today rather than at lot close, and the wage difference is small next to what a short shipment costs.

When should a piece be downgraded to B-grade instead of repaired?

Weigh whether the defect is repairable at all, how much operation time the repair would consume, and whether the repaired piece would pass the buyer final inspection. A skipped seam on a side panel is worth repairing; a needle hole in the middle of a plain front panel is not, because no repair survives inspection. Track the B-grade by defect type — stain, tear or hole, size issue, stitching defect — because the destinations genuinely differ: a stain may recover on a second cleaning attempt, a hole will not, and a size issue is a well-made garment that may sell as a second at full construction quality.

The test for whether your rework is under control is not the size of the repair pile. It is whether you can say, right now, how old the oldest piece in it is. If that question takes a walk to the checking table to answer, the pile is running the factory rather than the other way round.

Santosh Rijal is the founder of Scan ERP, a garment manufacturing ERP system designed for factory floor operations. He works directly with sewing lines, cutting rooms, and production supervisors across Nepal's garment manufacturing sector.