The call that started it
March 2024, Tuesday afternoon, and I'm at my desk in the quality office with two vendor scorecards open and a coffee going cold. The phone rings. It's a service shop customer — independent, four bays, been buying from us for maybe six years — and they've had three Valeo alternators come back in six weeks. Every one, they say, is bad.
Symptoms: headlights flickering at idle, battery warning light staying on after start, a faint squeal from the accessory belt. If you've ever looked up how to know alternator is going bad, that's the checklist — right down to the squeal. They wanted a warranty replacement and honestly, I could see their point. Three units, same symptoms, why wouldn't it be the parts?
My job was to figure out whether the parts were actually defective, or whether something else was going on. Ten years of doing this has taught me that when a customer comes in hot, the obvious answer is usually not the answer.
The logo that didn't look right
I started where I normally start — with the part itself. The shop sent over photos of the install, including the Valeo alternator wiring diagram they'd been working from. Except the pinout didn't match the housing. Their diagram had pin 1 as B+ and pin 2 as the indicator lead. Our reference shows the reverse for that part number. That alone would explain a charge light that never goes off.
But I wasn't ready to stop there. I wanted to verify the units were actually our units. So I pulled the decals. And this is where I caught myself doing something I've done a hundred times — squinting at the Valeo logo, comparing stroke weights, looking at the accent over the "e." It's not a fair test, really. I don't have a calibrated eye. But the logo on two of the three housings looked just a little heavier than it should — the kind of difference that makes you doubt yourself more than the part.
Turns out the logo was fine. The parts were genuine Valeo — but two of them had been re-manufactured and re-boxed by a third-party rebuilder somewhere upstream of our distributor. Not counterfeit. Just not new. Per FTC guidance on "rebuilt" versus "new" labeling, that distinction is supposed to be clearly disclosed, and it wasn't. That's a separate conversation I'm still having with the distributor.
The air filter on the other side of my desk
Here's where the story gets less tidy. While all that was going on, I had a separate ticket open — a repeat complaint about Levoit air filter units we'd resold through our small-appliance line. Same customer complaint twice in a month: weird smell coming from the unit after about two weeks of use.
Normally, that's user error. Filter saturates, unit runs too long, smell happens. But this customer had replaced the filter and the smell came back in three days. So I started asking dumb questions — where was the filter stored, how was it packaged, when did we receive it. Long story short: the pallets had been sitting next to a loading dock door that leaked when it rained. That's the smell. Not the filter. Not the customer. We replaced 43 units and wrote off roughly $8,200, give or take. (Note to self: get the loading dock door replaced before monsoon season.)
Different product, different complaint, same pattern. Someone — me, the customer, the distributor — assumed the failure was in the obvious place, when the actual failure was one or two steps upstream of it.
My pediatrician's differential
Same week, coincidentally, my youngest came home with a rash across his palms. Nothing scary-looking — small red spots, slightly raised. My wife took him in and the pediatrician spent about twenty minutes working through what she called a palm rash differential — basically, a mental checklist of every cause she could think of: contact dermatitis, hand-foot-mouth, fungal, allergic reaction to something he ate, a viral thing that resolves on its own.
She didn't guess. She eliminated. She worked down the list in order of how likely each one was and how cheap it was to rule out, and she sent us home with a plan. (He's fine, by the way. Contact dermatitis from a new soap.)
I sat in the waiting room and thought: why don't I do that with alternators? I know why. Because most quality complaints come with a story already attached — "your part failed." The story is usually wrong in the details. It's also usually right about the outcome — something did fail. Just not the thing anyone named.
What actually happened with the alternator
Back to the three units. Here's what the trace turned up.
- Two were re-manufactured units from an unauthorized rebuilder. Not counterfeit, but not new. That's on the distributor we bought from and we're still sorting out the credit.
- One was a genuine new Valeo unit and it tested clean on the bench. The failure was installation: the shop had used the wrong reference diagram for that specific part number, and two pins got crossed. Nothing wrong with our part. Everything wrong with the document they were working from.
We replaced the third one anyway, at no charge. Call it goodwill or call it a sunk cost, I don't know. The customer was understanding once I walked them through the trace. The other two got swapped by the distributor and we were credited.
In hindsight, I should have pulled the reference diagram before I pulled batch numbers. That would have saved me about ten days.
The part I keep coming back to
If I could redo one thing about that month, it's this — I would have built a differential first. I would have listed every plausible cause of "alternator fails" — bad part, bad install, bad reference, counterfeit, wrong part number, another chase entirely — and worked them in order of likelihood and cost. Instead I started with the most expensive check and worked backwards. Waste of a week.
I don't have hard data on how often that mistake happens across our industry. But based on the first-delivery rejections I review — roughly 60 to 70 a year — my sense is that maybe 30 or 40 percent of "product failures" are actually something else. Install, spec mismatch, storage, a bad reference document somewhere upstream. That's not a knock on customers. It's a knock on my own process for not asking the obvious question first: what else could explain this?
The fundamentals of quality review haven't changed in my lifetime — someone still has to catch the failure before it ships. What's changed is the number of places a failure can hide, which means the reference material on my desk matters as much as the parts on the pallet. Five years ago I would have gone straight to the part. In 2024, I go to the paper first.
What I changed
After that month, I wrote a one-page checklist for our desk. Before you accept a product complaint, list at least three possible causes. Not just the one the customer gave you. Work through the cheapest ones first. Sounds obvious when I type it out. It wasn't obvious in March.
We also started requiring every shop that buys from us to confirm the part number against our catalog — and we send the actual Valeo alternator wiring diagram for that specific part number with every shipment, not the generic one. That single change cut our alternator returns by roughly 40 percent over the next two quarters. I'd have to pull the exact numbers if you asked me for a hard figure.
And I started taking the pediatrician's approach at my desk. Before I call something a product failure, I ask what else could produce the same symptom. Turns out that's how you build a differential. And most of the time, the differential is the whole job.