The 9-Point Material Sourcing Checklist I Run Before Any PO Leaves My Desk
-
What this checklist covers, and who should use it
-
The nine checks, in the order I run them
-
1. Write down the end-use conditions before you open a single catalog
-
2. Ask for the actual test report, with the lot number on it
-
3. For carbon fiber, verify the number and the test method behind it
-
4. Check capacity concentration, not just the quote
-
5. Confirm mass per unit area and construction on the incoming roll, not the sample
-
6. Match the colorfastness test to how the product will actually be washed or exposed
-
7. Abrasion resistance — a cycle count without a test method is noise
-
8. Pattern alignment, stripe registration, and skew — the check that gets skipped
-
9. Build the total cost line before you compare two quotes
-
1. Write down the end-use conditions before you open a single catalog
-
Notes, and the mistakes I keep seeing
What this checklist covers, and who should use it
I buy two categories of material: composite reinforcements and technical textiles. Different supplier ecosystems, different testing regimes, same failure mode. Someone writes a spec, someone else approves a data sheet, and eight weeks later the lot lands and doesn't behave the way the spreadsheet promised.
I'm a procurement manager at a 320-person specialty converter. I've owned roughly $2.4 million in annual material spending for nine years, negotiated with 60-plus suppliers, and logged every PO in our ERP since 2016. The sequence below is what I actually run before a purchase order gets released. Call it 40 minutes on a first-time order, five on a reorder.
Since I formalized it in 2021, it's caught roughly $180,000 worth of problems before they became rework, reprints, and air-freight charges. Five minutes of verification beats five days of correction. That's not motivational poster material — it's arithmetic.
Use this if you're the person who signs the PO and gets asked why the material's wrong.
The nine checks, in the order I run them
1. Write down the end-use conditions before you open a single catalog
Not the product category. The conditions. Temperature range, UV exposure hours, abrasion contact, cleaning method, chemical contact, expected service life.
A nylon fishing jacket shell and a coated industrial fabric may both be "nylon," but one lives in salt spray and sun for 400 hours a season and the other lives indoors behind glass. Same polymer, completely different specifications. I've watched a buyer save 14% per yard on a shell fabric that failed DWR retention after five home launderings — the jacket ran in a coastal charter program, so every unit got washed weekly. The savings were gone by week six.
Write the conditions down. Make the supplier respond to those conditions in writing, not to the category name.
2. Ask for the actual test report, with the lot number on it
A data sheet is marketing with a logo on it. A test report has a laboratory name, an accreditation number, a sample date, a lot or batch identifier, and a method reference.
This is the check most people skip, in my experience. They accept "meets ASTM" or "passes industry standards" and move on. I want the PDF, and I want to confirm three things: the report is less than 12 months old, the sample traceability matches the lot being shipped, and the test method version is current.
I had a supplier send a colorfastness report from 2019 for a 2024 lot. They weren't being dishonest — the dye house had changed carriers and just never re-tested. That's a five-minute email. It's also a $6,000 problem if the order ships and fails.
3. For carbon fiber, verify the number and the test method behind it
Tensile properties get quoted loosely in this industry, and that's the fastest way to design a part that's 20% heavier than it needs to be — or 20% under-strength.
Two things to separate. First, the nominal figure. Toray's published data for T1100G carbon fiber lists a tensile strength of 6.6 GPa and a tensile modulus of 324 GPa. Those are real, published, verifiable numbers. Second, and this is what gets lost: they're nominal 0° strand values measured under a specific method on a specific specimen format. Your cured laminate in a quasi-isotropic layup will not hit 6.6 GPa in any direction. Design allowables live far below the headline figure, and they depend on fiber volume fraction, void content, cure cycle, and the environmental knockdown factors your engineers apply.
So when a supplier quotes a tensile number, I ask two questions: strand or laminate, and tested to which method? If they can't answer cleanly, the number is decoration.
The other half of this check is grade discipline. The full Toray range runs from T300 up through T1100, and the price differential between grades is significant. T1100G is the right call for a mass-critical aerospace or high-performance sporting application where the strength-to-weight trade justifies the premium. It's rarely the right call for a general industrial bracket that's stiffness-driven. I've seen T1100 specified where T700 would have passed every requirement, and the material line item alone went up 40%. Nobody asked, so nobody caught it.
4. Check capacity concentration, not just the quote
Pricing tells you what a supplier wants today. Market structure tells you what they can deliver in six months.
Publicly reported estimates have put Toray's share of the global carbon fiber market in the 30% range, with a considerably higher share in aerospace-grade fiber. I'd verify the current figure against a fresh market report before you put it in a business case — the number moves and the reporting lags. But the structural point doesn't change: a handful of producers control most of the world's carbon fiber capacity, and aerospace contracts get allocated first.
Practically, that means I ask every supplier two questions on the first order: which plant is this coming from, and what's your allocation priority if demand spikes? A supplier who can't name the plant is a trader, not a manufacturer. That's not disqualifying, but it changes how much buffer I build into the schedule.
5. Confirm mass per unit area and construction on the incoming roll, not the sample
Fabric weight is the single most common quiet substitution. It's easy to shave 8% of the gsm and almost impossible to see on a swatch.
I check mass per unit area against ISO 3801 on the actual delivered lot, and I keep a running log. Anything outside ±3% of the specified value triggers a conversation. I also check width tolerance, yarn count where it's specified, and — for wovens — pick count. If the delivered construction is 42 picks instead of 46, the hand feel changes and the tear strength changes with it.
Sound tedious? It is. It's also a 15-minute check that prevents a four-week dispute.
6. Match the colorfastness test to how the product will actually be washed or exposed
"Colorfast" by itself means nothing. Test conditions are the whole answer.
For home-laundered textiles, I reference the ISO 105-C06 series with the specific wash condition that matches the care label — the grading scale shift between a mild and severe wash condition is large enough to break a program. For outdoor exposure, AATCC 16 lightfastness with a defined AFU endpoint, plus a ΔE reading after exposure. For brand-critical colors, Pantone's own guidance puts the acceptable tolerance at Delta E under 2; between 2 and 4 it's visible to a trained observer, above 4 it's visible to most people.
Then I check the substrate. That same Pantone blue that matches beautifully on a 100% cotton poplin will read two full Delta E points off on a solution-dyed nylon. Suppliers know this. Ask anyway, in writing, and get the ΔE they're committing to on your substrate.
7. Abrasion resistance — a cycle count without a test method is noise
This one matters most on upholstery, and it's where the language gets sloppiest.
When a customer asks us to find the best upholstery fabric for dogs, the conversation isn't actually about a fabric name — it's about a testing strategy. Martindale and Wyzenbeek produce different numbers on the same fabric and they're not convertible. A 30,000-cycle Martindale result and a 30,000 double-rub Wyzenbeek result are not the same claim. I always state the method alongside the number, and I refuse to compare across methods.
For pet use specifically, abrasion is only half the test. I want: Martindale abrasion with the method and endpoint specified, snag resistance (a looped or textured weave that looks great in a showroom can be shredded by claws in a month), a cleanability check against the actual cleaning agents the end user will reach for, and — critically — an anti-soil or stain-release finish that survives cleaning rather than washing off after the third pass.
Anyone quoting a single "rubs" number with no method attached hasn't done the work.
8. Pattern alignment, stripe registration, and skew — the check that gets skipped
This is my "almost everyone misses it" step, so I'll spend a minute on it.
On outdoor ticking stripe fabric, the specification rarely says anything about the stripes beyond width and color. Then the cut-and-sew floor gets the roll, and the stripes are bowed three-eighths of an inch across the width. Two panels sewn together show a visible jog. Every panel gets re-cut. You've now paid for the fabric twice.
Three things go into my purchase order for any striped or patterned textile: maximum allowable bow and skew, measured across the full width; stripe pitch tolerance, in millimeters per repeat; and pattern repeat tolerance if it's a printed or jacquard design. Ask your cutter what they can actually work with and write that number into the PO. Don't assume the mill knows your tolerance.
The same logic applies to solid colors under different lighting. If your product gets installed outdoors under Denver sun, approve the color under daylight illumination, not in the mill's showroom with fluorescent overheads.
9. Build the total cost line before you compare two quotes
Quote price is one line item. The list I actually compare:
- Material price per unit, delivered, with freight terms specified
- Minimum order quantity and the carrying cost of the excess you didn't need
- Lead time, and the probability you'll pay expedite fees at least once
- Sample and testing fees — including whether the supplier's "free" testing is actually your testing
- Inspection and incoming verification cost on your side
- Failure cost × probability of failure
That last one is where the spreadsheet usually flips. A supplier 6% cheaper with a 15% higher chance of a lot rejection isn't cheaper — it's a lottery ticket with a negative expected value, and you're the one holding it.
I built this calculator after getting burned twice on hidden fees in 2022. It took four years and roughly 300 purchase orders to fully internalize this, but the lesson is short: the cheapest quote is a number, the cheapest order is a relationship plus a verified spec.
Notes, and the mistakes I keep seeing
To be fair to suppliers here — most of these problems are spec problems, not integrity problems. In my experience, when a lot fails, the purchase order was vague. Somebody assumed the mill knew what "outdoor grade" meant. Nobody wrote down the UV hours.
A few practical cautions:
Don't let the checklist become a bottleneck. A 40-minute review on a $400 reorder is waste. I run the full sequence on new materials and new suppliers; on reorders I only confirm lot-level test reports and the four key measurements.
Get the tolerances in writing, not the specifications. Everyone agrees on the nominal value. Disputes happen on the acceptable range. The PO should say ±3% on gsm and 300 DPI at final size on any printed component or packaging — with the pixel math checked against final trim, not the file's nominal dimensions.
Keep a sample of every approved lot. Labeled, dated, sealed. When a quality dispute starts in month five, an approved sample from the first shipment ends the conversation in ten minutes. Without it, you're arguing about memory.
Make one call before you make twelve emails. On a time-critical order, the written trail matters, but a five-minute phone call with the mill's QC lead resolves ambiguities faster than a week of documentation. Had a 36-hour window to lock a PO before a dye house changed its lot schedule last fall. Normally I'd run the full review. There was no time, so I did it by phone with the supplier I trusted most and confirmed the colorfastness report in writing afterward. In hindsight I should have pushed the delivery date instead. The eleven days until that first cut were not relaxing.
The checklist isn't a guarantee. It's just the cheapest insurance policy I've found — and unlike most insurance, you usually get the premium back in avoided rework.
