I’m a quality and brand compliance manager at a building products company. I review every article—OEM handles, decking boards, full door assemblies—before it goes out the door. Roughly 200 unique items a year. When I implemented our verification protocol in 2022, I thought the grind would get easier. It does, but not because suppliers get better. It gets easier because you get better at catching the subtle stuff.
It started with a crate of door handles
Last January, I found myself standing in a warehouse with 800 door handles spread across three pallets. I’d just measured the grip section on the fifth handle and it was 0.42 inches thick. Our spec called for 0.50 inches, with a tolerance of ±0.02. That’s not a slightly off part. That’s a different part.
We were building a run of Can-Am Defender doors for an outfitter. The exterior panels were spec’d in walnut elite woodgrain slatted composite cladding—a beautiful material that gives vehicle doors a warm, architectural look without the weight of real timber. The client had also chosen the matching walnut elite door handles with a finish we internally call “Black Corset Top”: a deep, dark walnut tone that visually tightens the slats. Think black tie, not black t-shirt.
Our brand is woodgrain. Everything we ship should look like it means it.
What I found in the pallet
The cladding itself was fine. The grain direction on the slats was consistent, and the spacing between slats was uniform. I ran the depth gauge across the emboss and it held within spec. That part made me feel better.
Then I opened the box with the handles.
The finish didn’t match. The walnut elite woodgrain slatted composite cladding was a warm brown, but the handles were matte gray-black. The woodgrain finish on the handles was, frankly, a decal. The pattern repeated every two inches. I could see the seam where the print had been laid over the plastic.
The heavier problem was geometry. The grip section measured 0.42” against a 0.50” spec. The countersink hole was too shallow, so a test screw sat 1.5mm proud of the surface. On a door handle, that means a customer will catch a fingernail on it every single time.
I flagged it to our vendor. Their first response was: “It’s within industry standard.” Maybe it is. But our contract isn’t with “the industry.” It’s with a spec sheet I wrote. And if we deliver a Can-Am Defender door with handles that look nothing like the approved sample, we own the problem. Per FTC guidelines (ftc.gov), claims about your product have to be truthful and substantiated. The claim on our invoice? “Walnut elite woodgrain finish.” The reality? Printed plastic. We weren’t going to send that out.
Rejecting the batch
I’ve been doing this role for over four years, reviewing roughly 200 unique items a year. I’ve rejected about 9% of first deliveries in 2024—mostly for small issues: off tolerances, inconsistent colors, packaging damage. Around 70% of those get fixed on the next attempt.
Rejecting this one wasn’t hard. The vendor did that for me.
I have mixed feelings about the rejection process. On one hand, I don’t like paying for unusable goods; it feels like a waste. On the other, I’ve seen what happens when a quality issue slips through. In 2022, a door handle with the wrong radius cost us a $30,000 redo and set back a project two weeks. The memory of that one makes every inspection feel worth it.
(Should mention: we had built in a three-day buffer before the outfitter’s install date, which saved us from a catastrophic delay.)
I sent the vendor a single email with three photos and the measurement data. I wrote: “I don’t care about industry standard. This doesn’t meet the product spec. Please redo these at your cost.” They didn’t love it, but they accepted it after a long conference call.
What the rejection actually cost
Rejections aren’t free. The direct costs were:
- Return freight: $2,100
- Overtime inspection of the replacement: $1,400
- Lost time at our finisher: $1,300
- Three days of schedule: priceless (or at least annoying)
If we had accepted the first batch, we would have saved $4,800 in the short term. But we would have shipped 400 doors with handles that looked like cheap prints, and then we would have dealt with the calls. The expected value of accepting them was terrible.
It’s the same lesson I see when buyers choose the lowest quote. People think expensive vendors are expensive because they overcharge. Actually, vendors who deliver consistent quality can charge more. Price is the effect, not the cause.
Lessons I keep in my toolbox
When a potential supplier comes in 15–20% under our current vendor, I run a simple checklist:
- Do they know our exact tolerance? If they quote to “industry standard” instead of our documents, they’re not quoting the product.
- Can they make a pre-production sample? A sample that matches the production run is worth 1,000 pages of certifications.
- What happens if it doesn’t match? A supplier who tries to argue why the part is okay is a supplier who will argue later too.
The second batch of door handles arrived eleven days later. It was perfect. The Black Corset Top finish had the same deep walnut character as the cladding, the grain line was continuous, and the grip section measured within spec. We assembled one on a test Can-Am Defender door and it looked like a factory option.
Looking back, a part is more than the number on a PO. USPS can deliver a first-class letter for $0.73 and a large envelope for $1.50, but they still measure thickness because even a millimeter over the limit changes the rate. In the same way, I measure thickness because a fraction of an inch changes the part. It’s not about being picky. It’s about whether the product is what we promised it would be.
The outfitter shipped those 400 doors on time and they’ve reordered twice since. I want to say the replacement handles cost us about $0.80 more per unit, but don’t quote me on that—it’s been a while, and I should check the ledger.
The point isn’t that I rejected a bad batch. The point is that the bad batch is the best thing that happened to our relationship with that vendor. After that, they knew we measured everything.
