A Quality Inspector's 5-Step Checklist for Verifying Custom Automotive Parts (Before You Pay)

I'm a quality and brand compliance manager at an automotive parts company. I review every outgoing part—roughly 200+ unique items annually—before they reach our customers. It's a role that's taught me a lot about where things slip through the cracks. And I'm not talking about the obvious stuff. I mean the kind of issues that turn a routine delivery into a $22,000 redo and a delayed launch.

A lot of the purchasing guides and tips you see online are written by people who buy components, not by the people who have to live with the results. This checklist is for manufacturing engineers and sourcing pros who need to verify a batch of stamped or machined parts, not just order them. If you're dealing with any custom work—stampings, dies, CNC parts, forgings, extrusions—this five-step process is the minimum I'd recommend before you write that final check. I've structured it in the order I follow during a physical inspection at the receiving dock.

Step 1: The Paperwork Audit (This Is Where You Catch 90% of Headaches)

Before you even touch a part, you need to look at the paperwork. I know, it's boring. But this is honestly where I find the most issues. A Certificate of Conformance (CoC) is not proof that the parts are good; it's a piece of paper. The real proof is the in-process inspection report.

Ask for the dimensional inspection report for the specific batch number you're receiving. A good supplier will give you this without pushback. A bad one will talk about 'industry standard tolerances' without showing you the data. I had a batch of 8,000 brackets a few years back where the supplier claimed a Cpk of 1.67 on the hole location. The actual report showed a Cpk of 0.89. The vendor had just grabbed the generic capability study from the tool tryout, not the production run. I rejected it, and the rework set us back three weeks.

Here's the key check: Is the lot number on the inspection report the same as the lot number on the box? Simple question, but you'd be surprised how often they don't match. Double-check the material certs, too. If you specified 6061-T6 aluminum and the cert says 6063-T5, that's an immediate red flag. I've seen a $500,000 project derailed because the wrong aluminum was used in extrusions. It looked right, but it didn't bend right.

Step 2: The First-Piece Visual (Look for the Signs of a Bad Die Change)

Once the documents check out, I pull the first part from a random box, not the top one. I grab one from the middle of a pallet. The parts on top might be clean for the photo, but the ones in the middle? That's where you see if the operator paid attention.

I look for two specific things: burr direction and surface finish inconsistencies. With stampings, a burr on the wrong side of the bend tells you the die clearance is off or the die is starting to wear. With a forging or an extrusion, I look for die-lock marks or flow lines that look abrupt. A small change in lubricant temperature can cause a surface defect that doesn't appear in any inspection report but will cause a failure in the field later.

What I'm looking for is consistency across the first five parts I pull. They don't all have to be perfect, but they have to be consistent. If one part has a burr on the z-axis bend and the next doesn't, that's a process control issue. It's not a shipping error; it's a manufacturing drift. I'd reject the batch right there and ask for a process capability review.

Step 3: Dimensional Validation (Don't Trust the Report, Trust the Caliper)

This is the most common mistake I see buyers make. They get a 14-page inspection report with all the critical dimensions checked off, and they sign the delivery. I always take my own measurements on three specific features: the locating hole, a critical-to-function surface, and the overall envelope.

I use a digital caliper and a height gauge. Not a CMM, just the basic tools. Why? Because the part has to fit in a fixture in an assembly line, not just on a coordinate measuring machine. The report might show the hole is +0.02mm, which is within spec, but if the burr on the edge of the hole is 0.08mm, it won't locate correctly. That burr is a non-standard defect; it doesn't show up on the standard inspection report for diameter.

I only believed this after ignoring it and paying for a $3,000 order that came back completely wrong. The report looked perfect—all pass. But the actual part didn't fit the jig. Since that incident, every contract I specify includes a requirement for 'functional gaging' on the locating features. A go/no-go gage reveals a defect an optical measurement machine misses.

Step 4: The Surface Finish Test (Trust Your Fingernail, Then Trust the Data)

I don't break out a profilometer for every check. That's overkill for a receiving inspection. But I do run two quick tests. First, I drag my fingernail across a critical sealing surface. If my nail catches a scratch, that part is a reject. It sounds low-tech, but it catches defects the report may not mention because they're from handling, not formation.

Second, I check the finish against a comparator block if I have one. A lot of CNC machined parts and extrusions from different vendors have specific Ra requirements (like 3.2 or 1.6 microns). The vendor might say 'within industry standard,' but the industry standard is defined in your purchase order, not by the guy in the factory. If the surface is too rough for a sealing application, you've got a warranty claim waiting to happen. A vendor who says 'within industry standard' for surface finish without showing you the Ra trace is hiding something.

Step 5: The 'Ignore the Clear Skies' Test (The Step Most People Skip)

This is the part of the list that I've learned the hard way. Most people stop after the first four steps. They see good papers, consistent parts, and valid dimensions, and they sign the PO. I've learned to ask one final question: What didn't you test?

I call this the 'ignore the clear skies' test. It's tempting to think the inspection is done. But the complex reality is that the supplier's report only covered the features you specified. Ask the quality manager what they do not test. For a heat-treated forging, it's often the case hard depth or the microstructure. They tested hardness (a quick check), but they didn't test the microstructure for retained austenite or decarburization. If your application involves high stress cycles, that's a question you need an answer to before you accept the part.

Every spreadsheet analysis pointed to accepting the batch. Something felt off. Turns out the supplier had skipped a hydrostatic test on a complex extrusion profile, and 1,500 units had micro-cracks. That cost us a field recall. It wasn't a spec failure; it was a scope assumption failure. Now I always ask for the test results for the properties that are hard to measure, not just the easy ones.

Common Mistakes to Avoid

I've seen a lot of parts come through the door, and the biggest mistakes are always the same. Skipping the paperwork audit is number one. People trust the supplier because they have a nice website. Don't. Trust the data. Not the handshake. The second is measuring only one part. You need a sample size. If you only check one part from a run of 50,000, you're gambling. A sample size of at least 10 from different manufacturing shifts gives you a baseline. The third mistake is ignoring surface finish. It's not about looks; it's about function. A mirror-finish part that's too smooth can be as bad as a rough one for certain welding or bonding applications.

The vendor who said, 'This isn't our strength for heat treatment—here's who does it better' earned my trust for everything else. I'd rather work with a specialist who knows their limits than a generalist who overpromises. Per ISO 9001:2015, Clause 8.4.2, if you're controlling an externally provided process, you need documented criteria for approval, not just a purchase order. Use that clause. It's your authority to ask for these reports.

If you walk into a supplier's plant and see they don't maintain a gage R&R schedule for their in-process checks, that's a bigger red flag than a single bad part. It tells you they don't trust their own data. And if they don't trust it, you shouldn't either.