Why Clothing Designs Fail in Production — And How to Catch It Early | AKFA
AKFA · 16 September 2026
Why your design won't survive production — and how to catch it early
Most designs that fail don't fail because they were bad designs. They fail because a drawing and a garment are different objects, and the gap between them isn't obvious until something comes back wrong.
We see the same handful of problems over and over. Every one of them is cheap to fix at the drawing stage, expensive to fix at the sample stage, and ruinous to fix after a production run.
Here are the six that catch people most often.
1. The fabric can't do what the design needs
This is the most common one, and it usually comes from a reasonable instinct.
Someone decides they want cotton. Cotton is soft, natural, breathable — a good choice for a tee or a sweatshirt. Then the same cotton gets specified for leggings, and leggings need stretch. A hundred per cent cotton has almost no recovery, so they bag at the knees within an hour of wear and never come back. What's needed is a cotton-rich jersey with a small percentage of elastane — usually around 95/5. It still looks and feels like cotton. It just behaves.
The same thinking catches people on padded garments. A body warmer or a gilet isn't made of a fabric — it has an outer shell, a lining and a wadding in between. Naming one fabric for it is like naming one ingredient for a sandwich.
What to do instead: describe what you want the garment to do, not just what it should be made of. "Soft, holds its shape, doesn't go shiny" tells a manufacturer more than a fibre name, and lets them come back with two or three options you can actually feel.
2. Your branding disappears into the fabric
Logos are designed flat, on a screen, at a size where every detail reads. Fabric is not flat.
Three things go wrong:
Print on stretch knit cracks. Ribbed cuffs, beanies, leggings — anything that stretches significantly in wear will distort and crack a printed graphic within a few washes. It looks perfect on the sample and tired by month two.
Embroidery sinks into pile. On velour, fleece, terry or anything with a raised surface, stitches settle down into the pile and fine detail vanishes. It's solved with a water-soluble topping over the fabric and a stabiliser underneath — but it does mean very fine lines and small text won't hold the way they do on a flat weave.
Chunky knits swallow thin strokes. A wordmark with elegant hairline letterforms will read beautifully at 200mm and turn to mush at 25mm on a 5-gauge knit.
What to do instead: send your artwork as a vector file and ask, before anything is made, what the minimum stroke width and text height are on your chosen fabric. Any manufacturer who has done this before can give you a number in millimetres. If your logo doesn't survive that number, a woven label patch will carry the detail that embroidery can't.
3. The pattern is random, but your logo isn't
This one is specific to patterned fabric, and it's easy to miss.
If a garment has an all-over pattern — camouflage, animal print, marble, anything organic — the pattern will vary from unit to unit. Most people understand that and are happy with it; it's part of the appeal.
What they don't follow through to is what it means for anything sitting on top of the pattern. If your logo is cream and the background is randomised between cream and brown, then on some units that logo will land on cream and effectively disappear. You won't find out until a box of them arrives.
What to do instead: ask for a solid block to be engineered into the pattern wherever branding sits, so the background behind your logo is fixed even while everything else varies. On knitted garments this is entirely standard. On printed fabric it's a placement question that has to be raised before the screens are made.
4. The fit was approved on one size
You get a sample. It fits. You approve it. The full run comes back and half the sizes are wrong.
Grading — scaling a pattern up and down across a size range — isn't arithmetic. Bodies don't change proportionally, so a pattern that's correct at medium and simply scaled by a fixed percentage will be wrong at XS and badly wrong at 3XL. Sleeve length, armhole depth and body width all change at different rates.
This is the single biggest driver of returns for new brands, and it's invisible until customers start sending things back.
What to do instead: have the pattern properly graded rather than scaled, and cut your first sample near the middle of your intended range rather than at either end. Grading outward from the centre is more accurate than stretching from one extreme, and the middle size is the one most likely to reveal proportion problems early.
Grading is a one-off cost. Once the graded pattern exists, every size costs the same to produce — there's no per-size premium on reorders. If you're planning a size range, send us the measurements and we'll price the grading.
5. Hardware has its own minimums and its own clock
Custom zip pullers. Branded cord tips. Moulded badges. Metal eyelets in a particular finish.
These are lovely details and they make a garment feel considered. They are also, frequently, the thing that delays an order by a month — because bespoke hardware needs tooling, tooling takes weeks, and hardware suppliers usually have their own minimum order quantities, often far higher than a garment minimum.
So you can end up committed to 1,000 custom zip pullers for a run of 50 jackets.
What to do instead: raise custom hardware at the very first conversation, not once the garment is signed off. There's almost always a middle route — off-the-shelf components in a custom finish — that gets you most of the way for a fraction of the cost and none of the delay.
6. "Sage green" is not a colour
Every colour word means something different to everyone who reads it. Worse, the same dye reads differently on different fabrics: a colour that looks rich on flat cotton can look washed out on a pile fabric, because the raised surface catches light differently.
What to do instead: give a Pantone reference wherever you can. Then ask for lab dips — small pieces of your actual fabric, dyed to your colour — before anything is cut. You're approving physical fabric rather than a colour on a screen that's calibrated differently from every other screen.
If you don't have a Pantone, send a physical object in the colour you want. A swatch, a garment, a paint chip. Anything real beats a description.
The pattern behind all six
Every one of these is a question that could have been asked before a single stitch was sewn.
That's really the point. None of these problems is exotic, and none requires you to be an expert. They require a manufacturer who asks the questions early rather than quoting what you sent and letting you discover the rest.
When you're speaking to a manufacturer, the useful signal isn't how quickly they say yes. It's whether they push back. A factory that takes "cotton leggings" at face value and quotes it will make you cotton leggings, and you'll find out what's wrong with them after you've paid.
Where this gets caught
The sample stage exists precisely for this. It's a real production piece — same fabric, same machines, same people who'll make your bulk order — and it's the last point where changes cost a revision rather than an entire run.
So use it properly. Wear the sample. Wash it. Stretch it. Look at the branding under a hard light. Tell your manufacturer what's wrong with it in detail, and don't approve anything you're not certain about. Changes at sample stage cost a week. Changes after production cost the order.
For a full walk-through of how the process works from first email to delivered garments, including real timings, see how it works: from first enquiry to delivered order.
Talk it through before you commit
If you're working on something and you're not sure whether it will make, send it over. We'd rather flag a problem at the drawing stage than discover it together at sample stage — and most of the brands we work with are making their first garment and aren't expected to know any of this in advance.