Optimal Is a Loaded Word: My 8-Point Checklist for Buying Custom Automotive Metal Parts

Why I keep a checklist

I've been handling stamping, die, CNC, forging, and extrusion orders for 12 years. I've personally made (and documented) 23 significant mistakes, totaling roughly $86,000 in wasted budget. Now I maintain our team's checklist to prevent someone else from repeating my errors.

This is for anyone who buys custom metal parts for a vehicle program but doesn't stand next to the press every day. It's not a replacement for PPAP, APQP, or FMEA. It's the stuff those systems miss when you're tired, in a hurry, or convinced the quote is already fine.

'Optimal' should come with a warning label

Our company name is Optimal, so I hear the word a lot. I've learned to treat it as a red flag. Search for 'optimal thermostat settings summer' and you'll get answers from 68°F to 72°F. Search for 'optimal temperature for sourdough starter' and someone will tell you 75°F, while another baker swears 80°F is better. Both can be true. Without a defined environment and measurement method, 'optimal' is just a hope.

In metal parts, a good spec doesn't say 'best possible.' It says target, tolerance, method, and who measures. If a drawing or supplier says 'optimal' in place of a number, ask for the range and the gage.

The checklist

Here's the list I use now. It started as a Post-it note after a bad day in 2017. The order matters: steps 1-3 take an hour, steps 4-6 require supplier input, and steps 7-8 take ten minutes but get skipped the most.

1. Write the decision criteria before you call anyone

You wouldn't set a thermostat without knowing whether your goal is comfort or electricity cost. Same logic applies to a stamped bracket. I get calls that start with 'I need a simple bracket.' That's like saying 'I need bread' without mentioning starter, flour, or bake time.

Define the function, annual volume, life cycle, corrosion environment, assembly method, appearance expectation, and cost target. Write them down. Then send that document to the supplier. Checkpoint: Can an engineer who has never met you tell exactly what to measure?

2. Find the 'flat plane crankshaft' in your part

Someone in your meeting is going to search 'what is a flat plane crankshaft' and get a short engine-history lesson. A flat plane crankshaft has crank pins arranged 180 degrees apart, which gives a different firing order and a different sound. The real lesson for a metal parts buyer is this: geometry dictates process. You don't stamp a crankshaft; you forge or machine it.

At Optimal, we handle stamping, dies, CNC, forging, and aluminum extrusions because each part needs the right process. A bracket might be stamped. A steering knuckle might be forged. A prototype might be CNC machined. If you call every part 'stamped,' you'll get the wrong process, the wrong quote, and probably the wrong material. Checkpoint: List what the part must survive—load, heat, corrosion, crash—and make sure the process matches.

3. Read the notes below the title block before comparing quotes

In my first year (2017), I compared three stamping quotes based on price. The cheapest wasn't cheaper; the drawing revision was old. The supplier had quoted the previous thickness and a looser tolerance. When the math was corrected, it wasn't the low quote anymore.

Read every note: material grade, thickness, surface finish, burr side, thread specification, and PPAP level. A 2.0mm bracket in SAE J403 1008 steel will quote differently than 1018. If the drawing only says 'steel,' you're comparing apples to oranges. Checkpoint: all quotes reference the same drawing revision and the same notes.

4. Verify the invisible specs (the air doctor filter problem)

Nobody types 'air doctor filter' into Google unless they're worried about fine particles in the air. On the plant floor, the same concern shows up as cleanliness on stamped parts. I once ordered 8,500 brackets without specifying cleanliness. They looked fine. The dimensional report was perfect. Then they failed a customer's cleanliness test because the oil residue was above the limit. That was a $2,100 rework plus six days of delay.

If your part will be welded, bonded, painted, or used near sensors, say that. Burr direction is another invisible spec. Check the first article's edges, not only the digital dimensions. And if the part is visible, ask for color and gloss data. Pantone's color tolerance guidance uses Delta E below 2 for brand-critical colors; automotive panels and trims use the same kind of measurement logic.

5. Ask for process capability, not just a sample

A sample is one point. A process is a distribution. You don't need to become a statistician, but you should ask for Cpk on critical dimensions. A Cpk above 1.33 tells me the process can stay inside the spec without constant heroics. A sample can fool you.

I learned this on a 5,000-piece order. The first article passed, but the second batch wasn't even close. The press operator had adjusted the process to make that first sample. That mistake cost us a sort, some angry phone calls, and a week of production. Honestly, I almost deleted this field from the checklist because it's embarrassing.

6. Ask what happens when the die breaks

The 'radiator damage repair cost' question usually comes from someone with a leaking radiator who wants one number. The answer is never one number because you have to decide repair or replace. For a stamped part, the same question applies to the die, but the dollar amounts are bigger. A customer once asked us to fix a transfer die after a coil misfeed. Repair cost was $4,700, but downtime was 11 days because the trim steel wasn't stocked as a spare.

Ask about wear items, backup tooling, and emergency repair response before you place the order. Also ask about 'standard turnaround.' Here's something vendors won't tell you: that number often includes buffer time for their production queue. It's not necessarily how long your job takes.

I once had to decide whether to approve a lower-priced transfer die from a new supplier. The upside was a 14% price reduction on a two-year program. The risk was a single-source tool that might not hold up. Calculated the worst case: re-tooling at $18,000 and a six-week delay. Best case: save about $6,500 per year. The expected value said go, but the downside felt too big. I approved it, then asked our engineer to build a backup trim insert. That compromise wasn't in the original plan.

7. Put packaging and logistics in the quote

This sounds boring, but boring saves money. I once saw 4,200 stamped housings arrive with flash rust because the purchase order said 'standard shipping' and didn't mention VCI paper. The supplier met the spec; my spec was incomplete. That was my fault, not theirs.

Define packaging, storage limits, Incoterms, and who owns the parts until they're received. If the truck is delayed, who holds the risk? A small paragraph in the RFQ saves a ton of pain later.

8. Create a 'what if' file

After the third rejection in Q1 2024, I created a pre-check list that now lives in every quote folder. It includes material certificates, first article photos, surface finish photos, cleanliness test results, and the email thread where we approved deviations.

Why? Because when something goes wrong, you need to answer 'what changed?' in 15 minutes, not three days. We've caught 47 potential errors using this checklist in the past 18 months. Not all would have caused a recall, but a few would have caused a reorder.

What I still get wrong

I still chase perfect numbers. Last summer, I spent a weekend adjusting every vent in my house because I was convinced there was one ideal setup. That was basically the same instinct that made me request a 0.01mm tolerance on a part that needed 0.05mm. It took a patient supplier to talk me down.

I also get tempted by lower tooling quotes. Sometimes that's fine. We've gotten a lot of value from overseas die builders when the engineering review was honest. But if a supplier says 'life-time guarantee on a die,' that's a red flag. Tools wear. Anyone who says otherwise isn't being straight with you.

The industry has changed. What was best practice in 2020 may not apply in 2025. Digital measurement data and simulation have transformed how we catch problems before steel hits the die. But the fundamentals—understand the function, verify the spec, plan for failure—haven't changed.

Bottom line: if you take one thing from this, make it step 3. Read the notes below the title block. It's the dullest step and the one that has saved me more money than all the others combined.