The sample room is a sales cost

The department nobody bills

Serapion's sample room is nine people: a pattern maker on CAD, a cutter, five machinists, a finisher and a coordinator. Its cost sits inside factory overhead. That means the cost is spread across every piece the factory ships. And that means nobody can tell you what any buyer costs to chase.

Hoda Meligy has run it for eleven years and has never been asked for a number. In February 2026 Hilmy asked her for two: what does a sample cost, and how many did we make last season for enquiries that produced nothing.

What a sample actually costs

Build it from the bottom, the way you would build a cost sheet.

The trouser's standard time in bulk is 21.5 minutes, spread across a whole line. A sample machinist sews the whole garment alone. So the same garment takes 126.0 minutes, which is 5.86 times the bulk figure. That one ratio is the most misunderstood number in a sample room. It is also why "we will just run a few off the line" is never true.

Fabric is the other surprise. The proto (the first trial garment) needs 1.9 metres. But the mill will not cut less than 30 metres for development. So USD 4.35 a metre becomes USD 130.50 on the invoice. The honest way to carry that cost is to spread it across all the samples that cut will serve.

Two things in that table are worth arguing about at your own factory.

Fabric is the largest line in a first proto, not sewing. Getting the development minimum cut down, or sharing one cut across three buyers' protos in the same base cloth, moves more money than anything you can do to the machinist.

Coordination is the second largest, and it is invisible. Reading the tech pack, chasing the trim, answering the comments, packing the courier. That is USD 12.04 that appears on no timesheet in most factories.

One season, by buyer

Serapion's autumn season: 47 enquiries, 214 samples, 9 orders won.

That is just under USD 320 an enquiry and USD 1,671.09 per order won.

And 168 of the 214 samples belong to enquiries that never became an order: USD 11,928.00, or 79.3% of everything the sample room made.

That is not waste, and calling it waste is the mistake

The instinct on seeing 79.3% is to cut it. Do not. Those samples are what being asked for a price costs you. A factory with no unsuccessful samples is a factory nobody is asking.

The right conclusion is different, and it changes behaviour: the sample room is a cost of sales, not a cost of production. The money is spent before an order exists, on the chance of one, exactly like a salesperson's travel. Accounting standards are clear that selling costs do not belong in the cost of the stock you make. They are treated as an expense of the period they happen in. Yet almost every garment factory buries them in a production overhead, where no buyer, no product and no salesperson can ever be charged with them.

Move the number, and three decisions appear that did not exist before.

The buyer who never places

Kirkwall took 38 samples across two seasons and placed nothing. That is USD 2,622.56. It is not a moral failing on Kirkwall's part. Buyers sample far more than they buy, and they always will. It is simply a decision Serapion had never made, because nobody had ever put the number in front of anyone.

The reflex is to charge for samples. Price the reflex before you reach for it.

Charging USD 90 a sample across the season would have recovered USD 19,260. Serapion's average contribution on an order won that season was USD 12,400, and it won 9 of 47 enquiries. So charging pays for itself only if fewer than 1.55 orders walk away. That is 8.11 enquiries, or 17.3% of them.

The number a buyer actually feels

None of the above is what the buyer experiences. The buyer experiences elapsed time, and Serapion measured that too. From tech pack received to courier collected, the average was 19 days. Only 3.4 of those days were work.

So 82.1% of a proto's lead time is queue.

That matters commercially far more than the USD 95.60. The queue is what makes a buyer place the development somewhere else. It is also the one part of the whole onboarding chain in the next lesson that the factory controls completely. A sample room that promises seven working days and means it is a sales asset. One that promises seven and delivers nineteen has trained its buyers to start earlier somewhere else.

Act as a cost accountant who has worked in garment factories and is sceptical of overhead pools. Help me cost my sample room from the bottom up and then split a season of it by buyer. My sample room: [PEOPLE BY ROLE, LOADED HOURLY COST FOR EACH ROLE]. For a typical first prototype of my main product: [PATTERN AND MARKER HOURS, CUTTING HOURS, SAMPLE SEWING MINUTES, THE SAME GARMENT'S BULK STANDARD MINUTES, FINISHING HOURS, COORDINATION HOURS, FABRIC METRES NEEDED, MILL MINIMUM DEVELOPMENT CUT IN METRES AND PRICE PER METRE, HOW MANY SAMPLES THAT CUT SERVES, TRIM MINIMUM PACK COST AND HOW MANY SAMPLES IT SERVES]. For a later round: [THE SAME FIELDS, WITH AMENDED PATTERN HOURS]. My last season: [FOR EACH BUYER — ENQUIRIES, FIRST PROTOTYPES, LATER ROUNDS, ORDERS WON, CONTRIBUTION EARNED ON THOSE ORDERS]. Also: average elapsed days from tech pack received to courier collected, and average working days actually spent on one. Do the following. First, give me a cost for a first prototype and for a later round, line by line, and tell me which line is largest. Most factories guess wrong. Second, give me the ratio of sample sewing minutes to bulk standard minutes and explain what it means for anyone proposing to run samples on a production line. Third, total the season by buyer and show sample cost per enquiry, per order won, and against contribution earned. Fourth, tell me what share of the spend belongs to enquiries that produced no order, and say explicitly that this is the cost of being quoted rather than waste. Fifth, name any buyer whose sampling is not paying for itself, and compute the walk-away rate at which charging for samples would break even. Sixth, tell me what share of my sample lead time is queue rather than work. Show the arithmetic.

What to take away

1. Cost a sample from the bottom. A sample machinist sewing a whole garment takes several times the bulk standard time, and the development fabric minimum is usually the largest line.
2. Total a season's samples and split them by buyer. Almost every factory finds that most of the spend belongs to enquiries that produced no order.
3. That spend is not waste. It is what being considered costs, and it should be classified as a cost of sales rather than buried in production overhead.
4. Once it is split by buyer, you can see the buyer who samples and never orders. Charge that one, keep the buyer who orders free, and measure.
5. Never price a blanket sample charge without estimating the enquiries it will cost you. Serapion's break-even is a walk-away rate it has never measured.
6. The buyer feels the queue, not the cost. Most of a proto's lead time is waiting, and it is the one delay in the whole onboarding chain that is entirely yours to fix.