The Shingle Batch That Passed Every Paper Check—and Still Failed

When the Backup Plan Became the Plan

By late 2023, our primary supplier for commercial roofing materials had become a problem. Not a dramatic one—the slow, grinding kind. On-time delivery was sliding from around 94% to somewhere under 80%, and every late shipment meant storage costs we hadn't budgeted and project schedules that had to be re-sequenced.

I'm the quality and brand compliance manager at Valor. We run two product lines—hearth products (gas fireplaces, fireplace inserts, wood stoves) and roofing materials (metal roofing, underlayment, membrane, asphalt shingles)—so most of my job is checking that what ships under our name actually matches what we say it is. In 2023, I reviewed over 200 unique SKUs. On the shingle side, roughly 12% of first deliveries got rejected, and that number was creeping up.

So when the main supplier started to wobble, we went looking for a second source. Not just a quote—an actual audit. Samples, spec sheets, third-party lab work.

One small tangent, because it comes up more than you'd think: searching for "Valor" in building materials will also surface iron Valor CrossFit equipment and a few unrelated companies. It's a reminder that in B2B procurement you verify the legal entity, not the name on the quote. We learned that the ordinary way—by almost wiring a deposit to the wrong vendor in 2021.

The Quote That Looked Right

The new supplier came in 9% under our incumbent. On paper, that's a line item I can defend: $94 a square versus $86, on a 480-square school reroof, works out to roughly $3,800 in material savings.

Honestly? It was tempting. But the real win wasn't the $3,800. It was having a second qualified source.

We'd learned that lesson in 2022, the hard way. A supply disruption on roofing membrane left us waiting six weeks; the expedited freight alone cost around $11,000, and we still handed over the project two weeks late. Single-sourcing isn't a strategy. It's an exposure.

So we ran the process. We requested samples. We sent them to an independent lab.

They passed.

What the Lab Actually Tested—and What It Didn't

This is where I want to slow down, because it's where a lot of purchase orders quietly go wrong.

The governing product standard for asphalt shingles in the U.S. is ASTM D3462—that's the standard specification for asphalt shingles made from glass felt and surfaced with mineral granules. Wind resistance is tested separately, under ASTM D7158 (the sealed-tab method) or ASTM D3161 (the fan-induced method), depending on which your project spec calls for. Impact resistance is its own thing, typically UL 2218. And installation requirements generally trace back to Section R905 of the International Residential Code, as adopted by your jurisdiction.

Here's the part that trips people up: "we meet ASTM standards" is not a specification. It's a marketing sentence. Which standard, at what threshold, verified by whom, at what frequency?

Our lab work on the samples covered granule mass, unit weight, dimensional tolerance, and wind resistance under both D7158 and D3161. Everything came back inside spec. The report looked clean.

What I should have written into the PO—and didn't—was batch-level verification. The samples were one thing. Five hundred squares shipped in March were another.

The Delivery on a Cold Monday

The shipment arrived the second week of March. Five pallets, staged in our Midwest DC where the ambient temperature was hovering just above freezing. Wrapper intact, labels correct, packing list matching the PO line for line.

I walked the pallets with our receiving lead, and nothing jumped out. So we did what we always did at that point in 2024: we pulled one pallet, opened three bundles, and eyeballed them.

They looked fine. Granule surface uniform, tabs aligned, no obvious delamination.

We almost signed off. That's the part I keep coming back to.

Instead, our receiving lead—who has a better nose for this than I do—suggested we pull a second pallet. And then a third. On the third pallet, bundle weights felt off. Not dramatically. Just enough that he flagged it.

We sent five random bundles to the lab for granule-mass testing against ASTM D3462. Three of the five came back below the minimum threshold. Not marginally. Clearly.

We pulled the whole shipment and sampled the remaining pallets. Of the 15 bundles we ultimately tested, five were out of spec on granule mass. The vendor's response was that the material was "within industry standard."

ASTM D3462 is not a suggestion. It's a specification. There's a difference, and that difference is what our receiving protocol had been quietly missing.

The Cost of the Cheaper Square

We rejected the shipment. The vendor redid it at their cost, which is what the contract allowed—but contracts don't recover schedule.

The reroof slipped 11 days. Liquidated damages on the project ran us around $6,200, and the re-handling—unloading, re-staging, re-shipping, rescheduling the install crew—added roughly $3,200 more. Call it $9,400, against a projected $3,800 savings.

I've written this math down more than once. In my experience, the lowest quote ends up costing more, in some form, more often than people want to believe. Not every time. But often enough that the base case should be skepticism, not optimism.

Here's the thing, though: I don't think the vendor was malicious. I think they had loose process control at the plant and a sales rep who genuinely believed their material met spec, because the paperwork said so. That's the failure mode most buyers don't account for. It's not fraud. It's drift.

What We Changed

Three things, and I'd argue they're the minimum for anyone sourcing shingles at volume.

Batch-level lab verification, written into the PO. Not samples. Samples are the vendor's best day. We now specify sampling methodology, test method (ASTM D3462 granule mass, among others), and the consequence of a failed result—return at vendor cost, credit within 15 days.

Five-pallet receiving inspection instead of one. Pull from the first, middle, and last pallets in the load; three bundles each. It adds about 40 minutes to a delivery and has caught two smaller issues since.

A qualified supplier list with a floor of two approved sources per critical SKU. We rebuild it annually. The vendor from this story is actually on it now—they reworked their process controls, passed a re-audit in August, and we've taken three shipments since without issue.

The Part I Still Wonder About

Looking back, I should have written batch-level verification into the PO before the first shipment, not after. At the time, the sample report looked solid, the timeline was tight, and asking for a heavier audit protocol would have added two weeks to onboarding. Given what I knew then, the call was defensible. Given what I know now, it was wrong.

I also have mixed feelings about the liquidated-damages piece. Part of me thinks we should have eaten the 11 days and kept the relationship friction low. Another part knows that a supplier who only feels the cost of a bad batch through conversation will treat the next bad batch as a conversation.

The compromise we landed on: liquidated damages stay in the contract, but we build in a documented-remediation path for a first occurrence. Fault-tolerance without fault-blindness. That's the balance I'd recommend to anyone running supplier quality in building materials.

One last note on standards, because it's easy to lose the thread here. All the ASTM documents I mentioned—D3462, D7158, D3161—are copyrighted and sold through astm.org, and the versions matter; specs get revised. The IRC requirements are published through the ICC and adopted state-by-state with amendments. If your compliance program is built on a PDF someone forwarded in 2019, it's probably built on the wrong revision.

The shingle batch that failed wasn't the cheapest one we could have bought. It was the one that looked like the cheapest one we could afford. That's the version of cheap that costs the most.