The Fabric Spec Problem: What 'Luxury Curtain Fabric' and 'PU Leather' Actually Get You

2026-11-20 · Jane Smith

The order that looked perfect on the quote

I've been handling fabric sourcing and production orders for nine years. In that time I've personally made — and written down — 11 significant specification mistakes, totaling roughly $41,000 in wasted budget. Our team's pre-production checklist exists mostly because of them.

The most expensive one started with a quote that looked completely fine.

February 2023. A hotel lobby refurbishment, 62 drapery panels, sun-exposed on two sides. The supplier's sample arrived, we matched it against the interior designer's reference, and it was close enough that nobody argued. The line item read luxury curtain fabric, 280 gsm, 100% polyester, FR treated. I signed it off.

Fourteen weeks after installation, the south-facing panels had gone from deep navy to a dusty gray-blue. Not a subtle shift — the client measured it with a handheld spectrophotometer and the Delta E came back above 4, which is visible to basically anyone walking past without any training at all. We replaced all 62 panels.

Fabric: $6,800. Air freight to hit the reinstatement date: $1,900. Rebooking the install crew: $740. Nine days of schedule slip across the wider project. Call it $9,400, plus a client who now sends every sample to an independent lab before she signs anything.

For about six months I was convinced the supplier had cheated us.

They hadn't. I'd done it to myself.

What I thought the problem was

My first instinct was that we'd bought from a bad mill. So I did what everyone does after a failure like that — I switched suppliers, added a 7% price buffer for "better quality," and moved on.

The second fade happened eight months later, on a different project, with a different mill. Different fabric weight, different finish, same outcome.

That's when I stopped blaming suppliers and started reading my own RFQs. What I found was uncomfortable: almost nothing on them was a specification. They were descriptions. Luxury. Premium hand feel. Hotel grade. High durability. Twelve words of mood, three numbers, and no test method anywhere on the page.

Here's the thing about a descriptor: a mill can read it honestly and still ship you something that fails. "Luxury curtain fabric" doesn't tell anyone which failure mode you're trying to prevent. It tells them what price tier to quote you.

Every fabric fails somewhere. The only job of a real spec is to name the failure you're willing to pay to prevent — and the test that proves it.

For sun-exposed contract drapery, the numbers that actually mattered were lightfastness to AATCC 16 (xenon arc) or ISO 105-B02. A Blue Wool rating of 4–5 is a common baseline for that exposure class, though it moves depending on the program and the latitude — I've worked to specs that demanded 6 and specs that shrugged at 4. I hadn't written either one down. I also hadn't specified dimensional stability after cleaning, or the flame standard (NFPA 701 in the US, EN 13773 in Europe), or seam slip at the heading.

We paid roughly $140 extra per panel to get lightfastness right on the replacement order. Against $9,400 in rework, that's a no-brainer in hindsight. It was not obvious at the time, because the cheap quote and the expensive quote looked identical in the only dimension I was measuring: price per meter.

The second trap: what is PU leather, actually?

Same problem, different vocabulary.

PU leather isn't leather. It's a polyurethane coating applied to a base fabric — usually a polyester knit or woven — sometimes with a release paper pressed into it to fake a grain. Bonded leather is a third thing entirely: shredded leather fiber glued into a sheet with a binder. Regulators in the US have gone after products marketed as leather that contained no meaningful leather content, which tells you how loose the labeling got before anyone pushed back.

So when a purchase order says "PU leather," it has said nothing useful. Two PU leathers at the same price per meter can differ by a factor of three on durability, and the difference is invisible in a sample book. What separates them:

That last one is the line item nobody puts in writing. Here's something suppliers won't volunteer: polyester-based PU coatings hydrolyze noticeably faster in warm, humid storage than polyether-based ones. Same look, same hand, same quote — different five-year outcome. If your PU leather sits in a bonded warehouse through a wet season, this is the difference between a product and a return.

We shipped 1,400 pieces of a PU-trimmed bag in 2021 that cracked at the fold line in month nine. Not month one. Month nine — long after the inspection window, long after the balance payment. That one cost $2,300 in credits and a retailer relationship I haven't fully recovered.

The phrase that actually does work

Not everything on a spec sheet is a mood word. Branded fiber is the exception, and it's worth understanding why.

When a fiber carries a brand — REPREVE is the one I deal with most — the fiber content stops being a claim and starts being something a third party can check. That matters enormously in recycled material, where "recycled" on a spec sheet has historically meant anything from 100% post-consumer bottle feedstock to a token percentage buried in a blend.

For repreve swimwear fabric, the practical value isn't the marketing story. It's that the fiber can be traced back through the brand's verification program to a specific production batch. That's the part that survives an audit, a customer questionnaire, or a green-claims review.

But — and I learned this the slow way — the brand name on your purchase order does not put the fiber in the box. I've seen a mill quote a branded recycled yarn and then substitute on the third bulk lot because the yarn was back-ordered and nobody wanted to delay shipment. The fix is boring: ask for the transaction certificate that links your specific order to the fiber, and make substitution without written approval a rejection condition. We put that clause in during 2022. We've used it twice.

One more thing on swim, since it's where I see the most misplaced blame. When a swimsuit fails in chlorinated water, the fiber that fails is almost never the polyester — it's the elastane. Recycled PET is chemically the same polymer as virgin PET; the difference sits upstream in the supply chain, not in the polymer backbone. So the performance conversation should be about the spinner, the texturing, the knit tightness, and the chlorine resistance testing (AATCC 162, or ISO 105-E03 for the European programs). Argue about those. The word "recycled" isn't a performance variable.

What this actually cost, over nine years

Eleven mistakes. Roughly $41,000. Three of them account for about 60% of that number, and every one of those three came from an RFQ where I'd written a feeling instead of a test.

The money is the easy part to report. The harder cost is the two years where our team didn't trust any quote I approved without re-checking it, and the client who now independently lab-tests everything we send her — which is fair, honestly, but it adds eleven days to every timeline we run for her.

The fix, which takes about ten minutes

I'm not going to stretch this part out, because the problem is the interesting bit.

Every RFQ we send now answers five questions before it mentions a price:

  1. Where does it fail? Sun, chlorine, abrasion, humidity, cleaning, flexing. Name the primary failure mode in the order it matters.
  2. Which test proves it? Method number and minimum result. Not "high," not "excellent." A number.
  3. What's allowed to differ from the sample? Color tolerance in Delta E units, weight tolerance in percent, hand feel described in measurable terms.
  4. Who verifies, and with what document? Lab report, transaction certificate, inspection record.
  5. What happens on substitution? In writing. Rejection, credit, or rework — pick one and put it in the PO.

That's it. No framework, no software. It fits on half a page.

I have mixed feelings about how long it took me to get here. On one hand, eleven mistakes over nine years is not a fast learner's record. On the other, I now sit in supplier meetings where I can tell within four minutes whether the person across the table knows what their own fabric does at year three — and that's a skill I only got by paying for it.

The quote is a number. The spec is the decision. Most people only find out which one they were actually negotiating about after the panels are already on the wall.

Leave a note