Why I’ll Never Skip Pre-Production Sourcing Checks Again — A Quality Manager’s Perspective

Opinion: The 12-Hour Sourcing check that costs less than a 48-hour recall

I’m gonna say something that might get me side-eyed by some procurement colleagues: If you’re not doing at least a 12-point pre-production sourcing audit on every metal stamping die before it hits the press line, you’re effectively gambling with your supplier’s reputation—and your customer’s trust.

I know, I know. “But we’ve worked with this vendor for years.” Or “The CAD model looked perfect.” I’ve heard every excuse in the book. And I’ve also seen the aftermath of skipping that check. Over the last four years, I’ve rejected roughly 12% of first deliveries from new tooling sources alone—not because the parts were bad, but because the spec was slightly off in ways that would only become visible after 5,000 cycles. And by then, you’ve already committed to production.

The contrast that changed my mind

I remember a specific run about two years ago. We had two different tooling sources quoting for a high-volume progressive die for a bracket that would eventually hold a Hella square fog light assembly. Source A had a slightly higher upfront cost—maybe $3,000 more per die—but their spec sheet included a detailed inspection plan with CMM reports for every critical dimension, plus a series of countersunk hole depth checks.

Source B was cheaper. Their quote was clean. Too clean.

When I compared the first-article inspection reports side by side, I finally understood why the details matter so much. Source A had flagged a 0.2 mm deviation on a hole diameter that was within nominal tolerance for the bracket alone—but would have caused a misalignment with the Hella square fog light housing. That one deviation would have led to a $22,000 rework and a delayed launch.

Source B? Their report was just the standard form. No notes. No callouts. Just “Pass.”

Edging the “assumption trap”

I assumed “same specifications” meant identical results across vendors. Didn’t verify. Turned out each had slightly different interpretations of what “parallel to reference surface within 0.1 mm” actually meant on the shop floor. One vendor read it as “check once on the first piece.” The other checked it after every 50 parts.

Learned never to assume the proof represents the final product after receiving a batch of water pump fountain nozzles that looked nothing like the 3D-printed prototype we approved. The plastic sample was beautiful. The stamped metal version? A mess. They’d changed the tool path without telling anyone because it “saved 0.3 seconds per cycle.”

Why I always check the “unsexy” parts

Most people focus on the big items—like taillight housings or main structural brackets. But I’ve found the most expensive failures come from the small, seemingly insignificant parts. Take the S90 throttle body mounting plate, for instance. It’s a simple blanking and forming part, right? Except if the hole pattern is off by even 0.3 mm, the throttle body won’t bolt on correctly, and you get a check engine light on a brand-new car.

I’ve also seen water pump fountain brackets from a different supplier that were made from a slightly thinner gauge steel than we specified. On paper, it still “worked.” In reality, after 10,000 thermal cycles, those brackets fatigued and cracked. That defect ruined over 8,000 units sitting in a warehouse. The root cause? Nobody checked the incoming metal coil thickness against the purchase order.

How do you know the alternator is bad without checking the bracket?

Here’s a scenario that happens more than you’d think: a customer complains that their alternator is bad. How do you know the alternator is bad? Sometimes it’s not the alternator—it’s the bracket that’s misaligned. And that bracket was stamped with a die that hadn’t been checked for datum shift in 18 months.

We had a case in Q1 2024 where a customer’s quality team flagged high vibration in a prototype alternator assembly. After three days of investigation—scope creep, A/B testing, swapping alternators—someone finally checked the die alignment. The metal stamping die for the bracket had worn slightly on one of the guide pins, shifting the hole pattern by 0.5 mm. That one misalignment was causing the alternator to sit 1 degree off-axis. Which created the vibration.

The vendor insisted it was “within industry standard.” I pulled the Pantone-level alignment tolerance we’d agreed on in the contract—Delta E under 2, basically—and pointed out that while the part might function alone, its interface with the alternator was compromised. We rejected the batch. They redid it at their cost. Now every contract includes a note about checking all mating surfaces, not just the part geometry.

The checklist that paid for itself

The 12-point inspection checklist I created after my third mistake has saved us an estimated $8,000 in potential rework costs over three years. It’s not fancy. It’s literally:

• Confirm incoming material gauge and temper match the PO
• Verify die-set parallelism within 0.1 mm across 12 points
• Check all locating pins and datum features for wear
• Run a 50-piece first-article at full cycle speed
• Measure hole positions relative to the carline coordinate system
• Cross-check against the mating part (e.g., throttle body, alternator bracket)

I’m not 100% sure every team needs all 12 points. Take this with a grain of salt: some operations might only need 8. But the principle holds—5 minutes of verification beats 5 days of correction. Every single time.

But what about the cost?

Yes, rigorous pre-production checks add time and money to the sourcing cycle. You’re paying for a quality inspector’s time, maybe a CMM machine setup, and often a few extra day’s wait. On a 50,000-unit annual order, spread across three years, that might add $0.02 per part.

But I’ve seen what happens when you skip it. A $1,500 die-set bought from a low-cost supplier without verification led to $18,000 in rework costs on a single job because the tool started galling after 2,000 cycles. The Hella square fog light housing project I mentioned earlier? The $3,000 extra on the die saved us $22,000 in launch delays.

Looking back, I should have built this checklist from day one

If I could redo that decision, I’d invest in better specifications upfront. But given what I knew then—nothing about the vendor’s interpretation quirks or their QC staff’s reading of the print—my choice was reasonable. I learned. And now I’m passing it on.

Do not assume “same spec” means same result. Verify. Measure. Reject early. Reject often. The 12-hour sourcing check that costs a bit more upfront will always beat the 48-hour emergency recall that costs three times as much and damages your reputation.

That’s my view, anyway. I’ve seen it play out a dozen times. The prevention side always wins.