The Day a 9% Cheaper Supplier Cost Us 3x More: A Metal Fabrication Purchasing Story

A procurement manager at a metal fabrication company walks through a year of supplier decisions — steel coil delays, EDM cutting copper wire failures, Bohler welding rod substitutions, and the recurring 304 stainless steel vs 316 debate — and explains why total cost of ownership beats unit price every time.

The Monday morning email chain that started it all

August 14, 2023. Monday. 07:42. I was staring at eleven quotes in Excel and colored the seventh one green. A steel coil supplier we'd never worked with, priced 9% under the second-cheapest option. Per kilo that was $0.38, which at our annual volume worked out to roughly $47,000 a year.

I typed "best price, consider trial" in the notes column and added it to the Monday procurement meeting.

At the time I'd just taken over the full procurement budget for our metal fabrication division — around $1.8 million a year. We run industrial fixtures and small steel structures for equipment OEMs and EPC contractors. Our core materials are steel coil, stainless plate, filler metals, and EDM cutting copper wire. Those four categories make up over 70% of our spend.

I'd been buying metal for six years and thought I knew how to compare quotes. What I didn't know yet was that the $47,000 I "saved" that quarter would come back with interest.

From steel coil to cutting copper wire, everything looked "cheaper"

Let's start with the steel coil. The new supplier was legitimate — we checked the heat numbers against the MTCs. The problem wasn't the material. It was the timing.

They promised four weeks. The first batch arrived 11 days late. The second, 19 days late. Their excuse was port congestion, which is fair, but on my side the shop floor was scheduled around fixed delivery dates. When the coil slipped, we had to cover with what was already on the rack — usually 304 stainless plate instead of 316.

Now, the 304 stainless steel vs 316 question is one of the oldest arguments on our floor. Clients on coastal projects spec 316/316L. The budget looks tight, and someone always asks whether 304 can sub in. On paper, 304 has good general corrosion resistance; 316 adds roughly 2% molybdenum, which helps against chloride pitting. For a salt-air environment, the answer is 316, no debate.

That fall we quietly swapped 304 onto about 20% of the bracket assemblies to protect the schedule. That decision was going to come back too, but I didn't have the headspace to think about it yet — I was busy firefighting another problem.

Our EDM cutting copper wire kept snapping.

We run two wire EDM machines with 0.25mm brass wire. That September the wire broke 6–8 times a shift. I blamed the operators first, then the water quality, then finally traced it to a bad batch. The new lot had a slightly larger diameter tolerance, and it kept catching at the flushing nozzle.

So I went price-hunting again. Four quotes, the cheapest was $5 less per spool. I almost signed off. Our machining supervisor asked if he could run a single spool first. Good thing he did — that wire broke even more, and each broken run wasted 3–4 meters of cut. At around $11/kg, plus downtime, we ended up paying roughly $40 more per spool in hidden cost.

That was the first small lesson: cheap per spool isn't cheap per usable meter.

The turning point: switching away from Bohler welding rods

The thing that really woke me up was on the welding side.

That autumn we were fabricating gusset plates for a structural steel client. We'd been running Bohler welding rods for years — I'll skip the exact diameter here, but they were stable, no surprises. In the October budget review, finance suggested swapping them out for a cheaper brand with "basically the same composition." My analysis showed a 6% saving, roughly $21,000 a year, and I agreed.

The first month the weld appearance looked fine. The second month, porosity showed up. Rework. The client's QC engineer called us about a drop in NDT pass rates on spot checks.

We ran a controlled comparison — same base metal, same current and voltage, same welders. With the Bohler rods, the porosity rate was 0.8%. With the cheaper rods, 3.4%. Four times higher.

I don't have hard data on industry-wide filler-metal failure rates — our procurement system doesn't capture it — but what I can say from that single project: a reworked weld joint ran about $12. Across roughly 2,000 joints, we ended up reworking around 52 of them. That's $624 in direct rework, before you count the one-week schedule slip.

Add it up across the year, and the "saving" was gone.

We switched back to Bohler welding rods in January 2024 and stopped running price experiments on filler metal. The exception is where a specific alloy is needed — for high-hardness punches and inserts we now stock Bohler N690 stainless steel, a cobalt-alloyed martensitic chromium steel running around 1.1% carbon. We tested it side by side against 440C and got roughly 60 HRC after heat treatment with noticeably better wear. Pricier per kilogram, cheaper per punch.

Honestly, I'm not sure why market perception of N690 varies so much between shops. My best guess is the heat treatment window is narrow, and whoever has stable process control on the HT side tends to think it's brilliant while everyone else thinks it's overpriced. If someone reading this has a better explanation, I'd like to hear it.

The rework rule I use now

In Q1 2024 I deleted the "unit price" column from our procurement sheet and replaced it with "cost per conforming part." The formula is:

Unit price + freight + late-fee allocation + (expected rework rate × rework cost per part) + inventory carrying cost.

We re-ran all of 2023 through that model. Three findings:

  1. For commodity steel coil, on-time delivery beats unit price by a wide margin. A supplier that hits four weeks — even at $18/ton more — came out ahead of the cheap one once you priced our shop-floor idle time.
  2. Small-dollar items like welding rods and cutting copper wire have outsized leverage through failure rates. A 6% saving evaporates the first time a joint is reworked or a spool of wire shreds mid-cut.
  3. 304 vs 316 isn't a pricing question, it's a spec question. If the drawing says 316, don't "optimize" your way to 304. Any "close enough" call shows up at inspection.

What I actually do now

Our current process:

  • Steel coil and plate: two qualified suppliers minimum, alternating POs, so no single point of failure.
  • Welding consumables: locked to Bohler rods. No cost experiments. The test-run cost of a switch is higher than any theoretical saving.
  • Cutting copper wire: scored on break rate, not spool price. Suppliers report monthly; exceed the threshold and we move on.
  • 304 vs 316 and other spec-critical choices: decided at design review. Purchasing doesn't get a vote, which is exactly how it should be.

I keep one 2023 quote sheet folded inside my notebook. Every time someone tells me a new supplier is "9% cheaper," I pull it out and read the numbers again.

Maybe that's over-cautious. Maybe. But on my ledger, the most expensive supplier has never been the one with the highest quote — it's the one with the lowest.

Takeshi Morita

Takeshi Morita

Takeshi Morita is an independent high-performance alloys analyst covering titanium sheet, plate, bar, tube, Inconel, nickel alloys, cobalt alloys, and corrosion-resistant superalloys. He applies ASTM B265 and B443 requirements while examining grade chemistry, interstitial control, heat-treatment condition, tensile strength, elongation, creep exposure, bend performance, corrosion mechanism, and service temperature. His engineering guides help aerospace, energy, chemical-processing, and marine teams compare material forms and specify alloys for demanding thermal or corrosive duty.