Inventory in a garment plant: rolls, cut orders and bonded stock

What does a garment factory count that a distributor does not?

A plant counts four things where a distributor counts one. A distributor counts saleable goods. A plant counts fabric that has not been cut, work in progress on the line, finished garments waiting on a ship date, and trim.

Trim is the one that catches people out. It is cheap, it arrives late, and it stops a shipment as completely as missing fabric does. A style is six thousand woven labels short and the order does not go, whatever the sewing floor has finished. Because trim is cheap it is also the category most likely to sit in an unlocked cupboard with nothing posted against it, which means the first time anyone learns the count was wrong is the morning packing needs it.

These four are not one number with four sub-totals. Adding them together produces a figure that means nothing, because each is measured in a different unit and each answers a different question.

A distributor's on-hand is one question with one answer. A plant's on-hand is four questions, and the system that works is the one that keeps them apart while still tracing a finished carton back to the roll it was cut from.

Why is the roll, not the metre, the unit that has to be tracked?

Because two rolls of the same colour code are not the same colour. Fabric is dyed in lots, and shade varies between lots by enough to show on the finished garment.

So the stock unit is the roll: a lot identity, a width, an actual measured length taken at inspection rather than the length printed on the supplier's ticket, and a location. Metres are an attribute of a roll, not a stock unit in their own right. A system that reports 4,200 metres of navy twill on hand cannot answer the only question the cutting room is asking, which is whether there is enough of one dye lot to cut a 900 metre order without shading it.

This is the same mechanism that lot and expiry traceability uses for food and pharmaceutical stock, with the date taken out and shade put in its place. The identity belongs to the physical unit and it survives every movement. FIFO, first in first out, is a cost-flow rule, and it settles what a roll costs when it is consumed. FEFO, first expired first out, is a picking rule for goods that expire, and twill does not expire. The two rules are independent of one another, and neither of them tells the cutter which roll to pull. That is a shade decision made against a specific order, and the system has to let it override the cost-flow rule without breaking the ledger.

How does cut-make-trim consumption change the stock model?

Stock leaves against a cut order, not against a sale. A bill of materials says what the style takes, and the cut order consumes that from named rolls and named trim lots.

The bill of materials gives a planned consumption from the marker. The cutting room gives an actual. The difference between those two numbers decides whether your costing is honest, and it is the number most plants never post, because the planned figure is what went into the quotation and nobody goes back to it. Post the actual and the gap has somewhere to live: a cut order that took 40 metres more shell fabric than the marker allowed becomes a visible fact with a date, a style and a supervisor attached, instead of a slow drift in gross margin that surfaces a quarter later with no cause anyone can name.

Wastage and end-of-roll remnants are stock, not rounding. A 14 metre remnant is either a stock unit with an identity, a location and a lot still attached to it, or it is a hole in the count. Most cutting rooms choose the hole. The remnant goes on a shelf, it gets used for samples or sleeve facings, and no movement is ever posted. Twelve months later the count is short and there is no way to trace why. Give the remnant a row and it stays traceable, and it stays usable, because somebody can find the matching lot when a repeat order arrives.

Work in progress is a state a distributor's stock never passes through. A bundle on the sewing line is not fabric any more and it is not a garment yet. It has a cut order, a quantity, a size ratio and a position in the line, and it is worth real money. Plants that do not model it find out at count time that the difference between what they cut and what they packed is sitting in trolleys, with no unit to count it in. Counting it properly is the same discipline that cycle counting instead of an annual shutdown applies to a warehouse, brought onto the production floor.

What happens to bonded material that has not been re-exported yet?

It sits on your floor and it is not yours to sell. Material imported under a duty suspension for re-export is physically identical to cleared material and legally different, so the difference has to live as a state on the record.

Bonded and cleared are two states of the same physical pack. If the system shows 1,200 metres of interlining on hand and 900 of those metres came in under suspension against a specific export order, the quantity available for a domestic sale is 300 and not 1,200. Any system that models this by location alone fails on the first morning somebody puts a bonded roll on the wrong rack, and racks get mixed in every plant that has ever run a rush order.

The operational shape is easy to describe even where the detail is not. Material comes in under suspension. It is consumed into cut orders. Those cut orders become garments, and those garments leave in shipments. At some point the plant has to show that what came in without duty was matched by what went out, and account for the difference, including wastage and anything that never left the country.

What that reconciliation looks like, what tolerance applies to wastage, what you owe on material that is never re-exported, and which authority wants which document, is a question for your customs broker and your own compliance team. We will not make that ruling and neither should any software vendor. A stock system's job is narrower and it is still the hard part: hold the state on every pack, carry it through the cut order into the finished garment and the shipment, and produce the reconciliation on demand, so that the people who do make the ruling work from a record rather than a spreadsheet rebuilt under pressure.

What does a brand buyer's social and traceability audit want to see?

It wants a chain that runs from a shipped carton back to the fabric lot, with a document at every link. Auditors rarely want a report you built for them. They want the records you already keep to agree.

Two different audits arrive under the same heading and only one of them is a stock question. The social audit looks at hours, wages, age verification and the condition of the building, and none of that lives in the stock ledger. It belongs to payroll and to HR, and a vendor who tells you an inventory module answers it is selling you something. The materials and traceability audit is the one your stock system either passes or fails.

That audit asks for a specific chain: which supplier declaration covers this fabric lot, which receipt brought it in, which cut orders consumed it, which cartons those garments went into, and which shipment left with them. Every link needs a reference and a date, and the chain has to reconcile read in either direction. A plant that keeps this across four spreadsheets and a chat thread can still pass, by putting a merchandiser on it for a week. A plant where receipts, cut orders and shipments are rows in one ledger exports it in an afternoon. The difference is not the audit result. It is the week.

The obvious way to keep those rows honest is to scan at receipt and at issue rather than to type roll numbers into a form at the end of a shift. We should be plain about this: there is no inventory handheld Scale can put in a buyer's hands today. Offline scanning is the architecture the build is written against rather than a shipped demo, and the shape of it is set out in the piece on barcode inventory in a Cambodian warehouse.

What does a factory need from a stock system?

A plant needs three or four things a general stock module does not give it: roll identity with shade, cut order consumption posted against actual, and a bonded state that survives being cut.

Purchasing, receiving, costing, the ledger and the standard reports are solved problems, and a general platform gives you all of them at the start along with accounting your finance team already recognises. What general platforms handle poorly is precisely what makes a garment plant a garment plant. Roll-level identity with shade tends to be bolted on as a free text field. Consumption is modelled as a quantity against a product rather than against named rolls. Bonded status is usually faked with a separate warehouse, and that is the model that breaks first.

Scale builds the system to the plant. The cutting room, the bonded reconciliation and the brand's traceability format are the requirement, and the rest is a modest amount of accounting. What we will not do is pretend a package covers the floor because it covers the invoice. The stock engine underneath is described on the inventory service page.

If you are evaluating anyone, including us, take one real order into the demo. Bring two dye lots of the same colour, a cut order that consumed more than the marker allowed, one bonded roll, and one style short of buttons. Then ask to see the shipped carton traced back to both rolls. Most demos stop at the first of those. The ones that do not are worth a second meeting, and if you want to run it against us, tell us what your plant makes.

Do we have to barcode every roll?

Every roll needs an identity. A barcode is the cheapest way to make that identity survive contact with a busy cutting room, but a printed roll ticket with a unique number that a person copies onto the issue slip works, as long as it is recorded at both receipt and issue. What does not work is identifying rolls by colour name and rack position. Scanning at the rack is the architecture our build is written against rather than a shipped demo, so treat the identity as the requirement and the scanner as the convenience.

Our merchandisers run everything in Excel and it mostly works. Is that a real problem?

It works until two people need the same file, or until the person who built it leaves. The specific failure to watch for is that the spreadsheet holds planned consumption from the marker and nothing holds actual, so costing drifts and nobody can see where. If your files already record actual consumption per cut order, you are in better shape than most, and the case for software is about time rather than accuracy.

We are in an SEZ at Bavet. Does bonded material change the software or just the paperwork?

It changes both, and the software part is the smaller one. The system needs a state on every pack, that state has to survive the cut into work in progress and out into a shipment, and it has to produce a reconciliation of what came in against what went out. What that reconciliation must contain, and who it goes to, is for your customs broker and your compliance team to determine, not your vendor.

How does a plant stock system connect to the accounts?

It connects through valuation and cost of goods sold. The stock system produces the value of what is on hand and the cost of what shipped, and the accounts consume those two figures. It does not file anything with anybody, and no stock system should be sold to you on the basis that it does.

We do full package, not CMT. Does that change the model?

It adds purchasing and it does not change the four states. Under full package you buy the fabric and trim yourself, so the supplier lots, the lead times and the money at risk are all yours, which makes roll-level identity more valuable rather than less. The cut order, the work in progress and the finished goods behave the same way in both models.