Fabric chooses the factory

The fabric decides where a garment can be made. The choice is usually taken in the opposite order: a maker is found first, through a recommendation, a search or an introduction, and the cloth is brought to them afterwards as a preference. The dependency runs the other way, and it runs firmly.

01A fabric arrives carrying a machine, a skill and a finish

A maker's willingness to source a fabric is only a starting point, and they are not being dishonest when they do. Buying cloth is a different question from making a good garment out of it. A fabric carries three requirements into whatever building it arrives at: a machine, a skill and a finishing capability. Any of the three can be missing, and only one of them is visible from outside.

The machine requirement is more specific than a sewing machine being present. The needle has to suit the cloth, in size and in the shape of its point, because a point that cuts is correct for one construction and destroys another. The feed has to suit it: cloth dragged along by teeth underneath only will travel at a different rate on the top layer than the bottom, which is why some garments need the material moved from above and below at once. Presser foot pressure, thread, and tension are all set for a weight and a construction. The stitch itself is a decision, since a straight lockstitch has limited stretch compared with suitable stretch-seam constructions, an overlock trims and encloses an edge as it sews, and a chainstitch behaves differently again when a thread breaks.

A machine set for one cloth and asked to sew another does not refuse. That is the whole problem. It sews, and it sews slightly badly, and on a single garment the badness is small enough to be pressed out, trimmed away or simply not noticed. It becomes visible when the same setting runs through an order, because then the small fault is repeated identically on every piece, and a fault that is on every piece stops looking like an accident and starts looking like the garment.

This is why a visit tells you less than it feels like it does. From the door, a factory is rows of machines, and rows of machines look interchangeable. What is not visible is how those machines are set, which of them are permanently rigged for one operation, what the operators have been sewing for the past year, and what happens to a garment after it leaves the sewing floor. A building can look entirely capable of a garment it has never made.

02The routes are separate trades, not settings on one machine

Take five garments a new brand might reasonably want in one collection. Heavy denim has to pass several layers of stiff cloth under a needle without the seam stalling or the thread breaking, on machines set up for that load, and then it has to be finished: washed, dried, treated, whatever gives the cloth its final colour and its hand, the tactile character it has when a customer picks it up. Fine shirting is the opposite discipline. Fine needles, dense stitching, small seam allowances that must not fray or show through, collars and cuffs fused and pressed to a consistent standard, and an operator whose eye is trained on millimetres.

Cut-and-sew jersey is knitted cloth cut into pieces and joined on machines whose seams stretch with the fabric; put it through the wrong machine and the seam breaks the first time the garment is pulled over a head. Fully fashioned knitwear uses panels knitted to shape and typically linked at the edges, which is a separate trade with separate equipment, and the person who joins those edges is doing work that has no equivalent anywhere on a sewing floor.

Weatherproof outerwear adds another route. Unpicking coated or laminated cloth can leave needle holes, and repairs must be assessed against the required performance. Stitched seams may need compatible seam tape or another tested sealing method. The material, bonding conditions and construction have to work together. A maker who can sew the shape still needs to demonstrate that the finished garment meets the intended weather-protection standard.

Those are five production routes. They are not five settings on one. The equipment differs, the training differs, the way a fault shows itself differs and the finishing differs. A maker who is genuinely excellent at one of them may be unable to do another well, and unable is the accurate word: not unwilling, and not merely more expensive, but without the machines or the hands.

03Handling is the part that does not appear on a list of machines

Equipment is the requirement that can be bought. The second one cannot. A fine, slippery cloth moves under the presser foot and marks where it has been held, so it has to be cut, pinned and fed in ways that come from having done it often. A machinist meeting it for the first time will produce a seam that ripples very slightly, and slight ripple down the front of a shirt is exactly the fault a customer sees without being able to name.

Knitted cloth carries a different piece of knowledge. It comes off its machine under tension and has to be allowed to settle before it is cut, or it relaxes afterwards and the garment grows into a shape nobody drew. A building that runs jersey continuously treats that as a step. A building that does not will cut the cloth as it arrives, and the sizes that come out will be correct on the cutting table and wrong on a hanger. Heavy cloth asks for the opposite kind of skill: physically controlling the layers so the seam does not shift, and holding a topstitch straight across a thickness that fights the machine. Topstitching is a visible skill, and it wanders when the hand doing it is unpractised.

The same specialisation reaches back into the pattern. Seam allowances, the way an armhole is drafted, how much ease is built in and where a garment is expected to stretch rather than be shaped are all route-specific decisions made before anything is cut. A pattern drafted with woven logic and handed to a knit route produces a garment that is technically correct and wears like a costume. It is worth knowing that a pattern cutter is as route-bound as a machinist, because a maker who is new to a cloth is usually new to it in both places at once.

04Finishing fixes the character the customer touches, and it often leaves the building

The third requirement is the one founders discover last. What a garment feels like in the hand and what colour it actually is are largely decided after it has been sewn. Washing, drying, tumbling, pressing and treating are where a stiff cloth becomes soft, where a flat colour gains depth or loses it, and where a garment either sits properly or looks like it has been folded for a long time. On some clothes the finish is most of the character, and the sewing is the part nobody notices.

Each route has its own version of this, and they are not the same trade. Denim goes to a laundry. Shirting depends on fusing and pressing done to a consistent standard on equipment built for it. Knitwear is steamed and blocked to its measurements, which is what makes a panel a shape rather than a piece of fabric. Sealed outerwear is finished at the seams and then tested. None of these is a step that a building can improvise because it has a steam iron.

Some finishing also happens somewhere else, which is the part that changes how a maker should be assessed. The question of who makes a garment often has two answers, and the second one is chosen by whoever has capacity at the time rather than by the brand. That is how one order comes back in two shades: the garments went to two finishers, or to the same one on two occasions, and nothing in the arrangement required them to match. Asking what is finished inside the building and what leaves it is not an administrative question. It is asking who is actually deciding what the clothes look like.

05The cloth decision belongs early enough to name the maker

So the material decision needs to be made early enough to inform where the clothes get made. Once the cloth is settled, the question of who should make it has a shape to it. It names the machinery, the finishing and the kind of experience worth looking for, and it turns a vague search into a short list. Taken in the reverse order, the cloth gets quietly adjusted until it suits the maker who was already chosen, and the adjustment is offered as a technical improvement.

Having the cloth first also changes the conversation with a maker from one that invites agreement to one that can be checked. It replaces a general question with specific ones about a needle, a feed, a wash and a finishing arrangement, and specific questions are answerable. It gives a maker the material to say no usefully, which is worth considerably more than a yes: a maker who says that a particular cloth is not what the building runs, and names who does run it, has told you something true and saved an order. That answer is not available to somebody who has only been shown a drawing.

The drift in the other direction is hard to see because each step is reasonable. A lighter weight is suggested because it runs better on the line. A different construction is proposed because it is easier to get. A simpler finish is offered because it is done in house rather than sent out. Every one of those is sound advice from where it is given, and the sum of them is a different garment from the one that was designed. The symptom to watch for is a cloth specification that keeps changing while the reasons for changing it are always practical and never about the clothes.

06A sample proves the sample, not the order

The order also changes what a sample is worth. A sample made by a maker who does not routinely handle the fabric proves less than it appears to. Sample rooms are staffed by the most skilled people in the building, working one garment at a time, without a line running to output behind them. Given attention and enough attempts, a good machinist can produce a convincing garment in a cloth the factory floor has never run. The sample is real. It is not evidence of what happens when the same garment is cut in bulk and sewn at speed.

The conditions differ in ways that are easy to list once they are pointed out. One person makes a whole sample; on a line, a garment passes through many hands, each doing one operation repeatedly, and the garment is only as good as the weakest of those operations. A sample can be remade until it is right, and the attempts that failed are not shown. Sample cloth is often a short length bought for the purpose rather than the batch the order will be cut from. A sample is pressed by hand, with time; production is pressed by machine, to a cycle. Everything about the room is designed to produce one good garment, which is what it does.

This is one reason a good sample can fail to reproduce in production. Not deception, and usually not carelessness. The garment was made once, under conditions the order will not repeat. In bulk the seams pucker on machines that are not tensioned for the cloth. The knit grows, because nobody relaxed it before cutting. The topstitch wanders, because the operator has not sewn that weight all year. The wash comes back uneven, because finishing left the building and went to whoever had capacity. Each of those is a material meeting a route that was selected without reference to it.

Two more show up specifically at scale. Cutting a single garment means one layer under a hand; cutting an order means many layers at once, and a cloth that shifts or compresses in a stack comes out with pieces that do not match the pattern they were cut from. And colour continuity becomes a live question, because a sample was one length of cloth from one batch, while an order may be several, and cloth dyed at different times matches only if somebody has made matching a requirement. A garment can be sewn perfectly and still arrive as two slightly different products in the same box.

07Ask what a maker runs, not what they can do

The way to read a maker, then, is to ask what they run, not what they can do. Whether a factory can make something invites a yes and receives one. What is on the floor this week is a different question, and so are these. Which of these fabrics do they buy regularly, and which would they be sourcing for the first time. What is finished inside the building and what leaves it to be finished somewhere else. Which garments are hanging in the sample room that were made in a comparable cloth for someone else. Ask to see something they produced in the material as part of an order, not as a sample, and look at the machines while you are there.

Some makers genuinely do span routes, and the argument here is not that they cannot. It is that spanning routes is a claim to be verified like any other, and it is verifiable. A building running two routes properly tends to run them as two lines rather than one line reconfigured, with their own machines standing set up rather than changed over between orders. Ask which route the building was originally built around and which was added, because the added one is where the risk sits. Ask who runs it and how long they have been doing that work. Ask to see bulk from both routes, from orders rather than sample rails. Ask what leaves the building for finishing on each. A maker who really does span routes answers all of that in detail and usually volunteers the boundary themselves, because knowing where their capability ends is part of what makes them good at it.

The same logic governs a change of cloth later. A substitution that reads as like for like on paper, a heavier twill, a different knit construction, a blend with more recovery, can move a garment onto another route without anyone announcing it. If the material changes in kind, the question of who makes it reopens, even when the drawing has not changed at all. That is worth knowing before a mid development swap is agreed on a call.

One consequence is worth taking early. A first range that spans several fabric families may need several makers, several relationships and several sets of learning at once, before the brand knows which of its clothes people actually want. What that costs is not only attention. Goods arrive from different places at different times, so the launch cannot ship until the slowest of them lands, and a customer who buys two things gets one parcel late or two parcels. Labelling, folding and packing are done to each building's own habits unless somebody specifies otherwise, so garments that are meant to look like one brand arrive looking like two. Quality is judged against a different baseline in each place. And the founder is holding several relationships at the moment the brand most needs one clear judgement about its own clothes. Choosing cloth with the production route in view tends to pull a range together on its own, and that is a gain, not a restriction.

The strongest objection to all of this is that a good factory can make anything, and it is not a foolish objection. Skilled people do adapt, equipment is bought, and a serious maker learning a new cloth will often make a better job of it than a careless one who has run it for years. But the objection quietly swaps capability for competence. What makes a route reliable is not that the people are good, it is that the machines are already set for it, the hands have repeated it, the pattern conventions are known and the finishing is arranged. Those are properties of doing the work regularly, not of being good in general. A good factory is in fact the one most likely to tell you which routes it runs and which it does not, because that answer costs it an order and saves it a reputation. And a first production run is the worst available occasion for anyone to learn a route, because the learning gets paid for in the garments that were supposed to open the brand.

Decide the material, then decide whose hands the clothes belong in. A maker chosen before the cloth is an answer to a question nobody has asked yet.