Introduction
A lot of developers end up in apparel. Maybe it's merch for your open-source project, a side-hustle clothing label or a client who sells ethnic wear online. Whatever the reason, the first fabric order is usually where things go wrong.
The mistakes look familiar if you've ever shipped software without a spec. Vague requirements ("soft, good quality cotton"), no acceptance criteria, no versioning and surprises in production. The fix is the same too: write it down, define what "done" means and test before you scale.
Here's how to source fabric the way you'd build a feature.
1. Write requirements, not vibes
"Soft" and "premium" are not requirements. A supplier will happily agree with both and send you whatever they have in stock.
A usable fabric spec includes:
| Field | Example |
|---|---|
| Fabric type | Viscose georgette |
| Composition | 100% viscose |
| Weight | 70 GSM ± 5% |
| Width | 44 inches |
| Colour | Matched to approved swatch, lot noted |
| Shrinkage | Max 3% length, 3% width after one wash |
| Finish | Pre-washed, soft finish |
| Quantity | 40 metres |
Each line is something you can check. If you can't verify it, it isn't a requirement yet.
2. Always include tolerances
In code, you'd never write assert weight == 70. Real materials vary, so specify ranges.
- GSM: usually ±5% is reasonable
- Width: ±1 inch is common
- Shrinkage: set a maximum, not a target
- Colour: agree on how close counts as a match, ideally against a physical swatch
A spec without tolerances can't be failed, which means it can't really be enforced either.
3. Prototype before you scale
You wouldn't deploy to production without staging. Don't order 300 metres without a sample.
Order 1 to 2 metres first and run your own acceptance tests:
[ ] Measure 25 x 25 cm, wash as customers will, measure again
[ ] Check colour in daylight and indoor light
[ ] Press with the iron settings your customers will likely use
[ ] Sew one seam and pull-test it
[ ] Make one finished piece and wear it for a day
Plain everyday fabrics are a good starting point for learning this. Something like viscose georgette is a classic example. It's popular for kurtas and dresses, but it shrinks and loses strength when wet, so a wash test catches problems a hand-feel check won't.
4. Version your inventory: dye lots
Fabric has versions. Every dye batch, called a lot, can be slightly different in shade, even with the same recipe. If you mix lots in one production run, sleeves and bodies may not match.
Treat lots like release tags:
- Record the lot number for every roll you buy
- Keep a small swatch from each lot, labelled
- Cut all pieces of one garment from the same lot
- When reordering, send your saved swatch and ask for the closest match
This is the fabric version of "it worked on my machine." It worked on the last lot.
5. Mind the minimum order quantity
Mills often have high MOQs, sometimes hundreds of metres or kilograms per colour. For a first drop, that's like provisioning infrastructure for a million users before you have ten.
Options for smaller runs:
- Stockists and wholesalers that sell existing stock in smaller quantities
- Fabric markets, where you can buy by the metre and see the fabric in person
- Suppliers that let you mix colours within one order
Starting with wholesale fabric suppliers that sell smaller quantities lets you validate demand before committing to mill-level volumes.
6. Document everything, like a changelog
Keep a simple record for every fabric you use:
fabric: viscose georgette
supplier: <name>
lot: <lot number>
gsm_quoted: 70
gsm_measured: 68
shrinkage_length: 2.5%
shrinkage_width: 2.0%
wash_test: cold, gentle cycle, dry flat
notes: slight colour shift under warm light
date: 2026-11-13
Six months later, when a customer asks why their reorder looks different, you'll have an answer instead of a guess.
7. Get the spec in writing
Before you pay for bulk, send the spec to your supplier and get confirmation in writing: email, a WhatsApp message, an invoice note, anything that's timestamped.
It's your contract. If the delivery doesn't match, you have something concrete to point to. If the supplier won't confirm the basics, like composition, GSM and width, treat that as a red flag.
8. Know your fabric basics
You don't need to become a textile engineer, but a few basics help you write better specs:
- Woven vs knit: wovens (shirting, georgette, cotton) are stable. Knits (jersey, fleece) stretch and behave differently when sewing.
- Natural vs synthetic: cotton and viscose breathe and absorb moisture. Polyester resists creasing and shrinks less.
- GSM: higher usually means heavier and more opaque, but weave and fibre matter too.
- Width: affects how many pieces fit across the fabric, and therefore how much you need to buy.
For basics like everyday cotton, you'll notice these differences quickly once you compare a few weights side by side.
9. Plan for failures
Even with good specs, things go wrong. Build in buffers:
- Order 5 to 10% extra fabric for cutting waste and defects
- Inspect rolls on arrival before cutting, not after
- Keep a backup supplier for your core fabrics
- Don't schedule a launch the day after fabric arrives
The short version
1. Write a spec with measurable fields
2. Add tolerances
3. Sample and test before bulk
4. Track dye lots like versions
5. Start with suppliers that allow small quantities
6. Document results
7. Confirm everything in writing
8. Learn the basics
9. Build in buffers
If you can write a decent API spec, you can write a decent fabric spec. The material is different, but the discipline is exactly the same.

Top comments (1)
The dye-lot-as-release-tag idea is a great way to explain it. One addition from the buying side: ask for a lab dip or strike-off from the actual production lot before bulk approval, and write the shrinkage test method (wash temperature, drying method) into the spec so both sides measure the same thing. Otherwise the supplier's pass and your fail can both be honest results. I'd also have the supplier confirm in writing that the swatch you approved and the roll you get are from the same lot.