-
April 2023: the rework report I didn't want to read
-
How we got there: six years of comparing dollars per kilogram
-
The uncomfortable part: it wasn't the adhesive's fault
-
The sourcing project: six vendors, ninety days
-
The email that changed how I think about suppliers
-
What actually moved the numbers
-
One thing I got wrong about feedstock costs
-
What I'd tell the version of me from 2022
April 2023: the rework report I didn't want to read
I still have the printout. Forty-one bonded enclosure panels, flagged on a Tuesday morning, pulled off a customer's assembly line in Indiana. Bond-line voids on roughly a third of them. The rest were fine, which was somehow worse, because it meant nobody caught it before the panels shipped.
I'm the procurement manager at a 210-person manufacturer in Ohio. We build bonded metal-and-composite enclosures for industrial equipment — the kind of thing that gets bolted onto a machine and forgotten about for a decade. My budget covers adhesives, sealants, and specialty polymer products: about $380,000 a year across four product lines. I'd been running that category for six years. I knew every invoice.
What I didn't know, until that report landed, was that I'd been buying the wrong thing for most of those six years.
How we got there: six years of comparing dollars per kilogram
Here's the honest version of our sourcing process before April 2023. We ran a bid every eighteen months. Three or four distributors submitted quotes. I built a spreadsheet, sorted by unit price, and picked the lowest number that cleared our engineering spec. That was it. That was the whole system.
And it worked — for a while. Unit price dropped 11% between 2019 and 2022. I put that on a slide. My director liked the slide.
What the slide didn't show: the incoming inspection hours we'd added because lot-to-lot viscosity had started drifting. Three lots out of twenty-two in 2022 came in outside our working range, and one of those went to the line anyway because nobody wanted to stop a run over a number on a certificate of analysis. We'd also stopped getting meaningful technical support somewhere around 2021, when our incumbent distributor restructured its territory coverage and our rep became a name on an email signature.
I remember thinking, at the time, that technical support was a nice-to-have. Something I'd push for when there was leverage.
Then forty-one panels came back.
Total cost to make it right: $18,400. Labor for disassembly and re-bond, scrap on nine panels that couldn't be salvaged, expedited freight both directions, and a credit we had to issue to keep the account. Against a category where our entire annual unit-price savings in 2022 was roughly $9,000. So the failure ate two years of my wins in a single quarter.
The uncomfortable part: it wasn't the adhesive's fault
We ran a failure analysis. The adhesive met spec on paper. What failed was the interface — our surface prep had drifted when we changed a cleaning solvent earlier that year, and the substrate had changed too, because our aluminum supplier started shipping a different mill finish. Two small changes, neither one routed through me or through engineering, and the bond chemistry that used to work didn't anymore.
I didn't fully understand how adhesives work until that analysis. I'd been treating them like fasteners — buy the strength rating, install, done. But an adhesive doesn't have a strength. A joint has a strength, and the joint is adhesive plus surface plus prep plus cure profile plus the stress it actually sees in service. Change any of those and the number on the datasheet stops describing your reality.
The technical rep we eventually brought in put it plainly: "You're not buying glue. You're buying a process, and the glue is one input to it."
That sentence cost me $18,400 to learn.
The sourcing project: six vendors, ninety days
We rebuilt the category from scratch starting in June 2023. Six vendors, three months, and a scoring model that looked nothing like my old spreadsheet.
The new model had four buckets. Unit cost was one of them, weighted at 25%. The other three: technical support depth, supply reliability and documentation, and batch consistency history. I'll be honest — I made up the weights the first time, then adjusted them after talking to our process engineers. There's no standard formula. Ours is probably wrong for your plant.
The documentation bucket turned out to matter more than I expected. We're buying regulated chemistry, and every one of those products arrives with a safety data sheet, a certificate of analysis, and a chain of regulatory obligations attached. When I started asking vendors for lot-level traceability and current SDS revisions in the same format across all four product lines, half of them couldn't do it without a project plan. That told me something.
The supply reliability piece took longer. I pulled our receiving data for three years and mapped late deliveries, partial shipments, and short-dated material by vendor. What I found wasn't dramatic — nobody was catastrophically late — but the spread between best and worst was about eleven days on average lead time, and eleven days is the difference between running the line and renting a truck.
The email that changed how I think about suppliers
Here's where the story turns, and it's not the turn I expected.
One of the six vendors was a distributor carrying Huntsman's advanced materials line, among others. Their technical rep spent about two hours on our floor in July, looked at our high-temperature application on line three, and then sent me an email I've kept.
He wrote that his organization wasn't the right fit for that particular high-temp joint. The service temperature we needed sat outside where their structural epoxy range made sense, and he'd rather tell me that now than have us find out in eighteen months. He named the class of product we should be looking at instead, and then said — for the other three lines, here's what we'd recommend, and here's why.
I sat with that email for a while.
Every other vendor in the process had told me they could handle all four lines. Some of them probably believed it. But the one who drew a boundary around their own competence earned my trust on everything inside it. We ended up standardizing on a Huntsman Araldite structural epoxy for the aluminum-to-composite joints on two lines — the grade their rep specified after looking at our actual prep process, not the generic one from the catalog. For the high-temp line, we went elsewhere.
I've come to think that's the whole game. A supplier who tells you where they stop is telling you their map is accurate. A supplier who says yes to everything is telling you they haven't checked.
What actually moved the numbers
Our blended unit price went up 8% after the switch. I had to defend that in a budget review, which was not a fun morning.
What went down: incoming inspection time on adhesive lots, which we cut because the batch consistency data was there. Line stoppages attributable to bonding, which dropped from four in 2022 to zero in the fourteen months after the switch. Rework. Scrap. And the two engineer-days a month our process team was spending chasing down formulation questions, which nobody had ever costed out before.
If I add it up conservatively — and I've gone back and forth on these figures, so take them as directional — the category costs us roughly 4% less per year than it did under the "cheaper" supplier. On $380,000, that's about $15,000. Not a headline number. But it's real, and it doesn't come with a rework report attached.
One thing I got wrong about feedstock costs
I also learned to watch upstream chemistry, which I'd ignored for years. A lot of what we buy — polyurethane systems, solvent-borne products, certain resin intermediates — traces back to methanol somewhere in the chain. Industry estimates as of 2024 put global methanol production somewhere north of 110 million metric tons a year, with China accounting for roughly 60% of global capacity. That concentration means a capacity decision or an energy policy shift in one region can show up as a price move on my purchase order eighteen months later.
I'm not a market analyst and I won't pretend to forecast it. But I now put a methanol price line on my quarterly review, and when it moves, I ask our suppliers what it means for the next two quarters instead of finding out at renewal. That's the whole discipline. Ask earlier.
What I'd tell the version of me from 2022
Three things, and then I'll stop.
First: unit price is a data point, not a decision. Total cost of ownership in specialty chemicals includes incoming inspection, line stoppages, rework, regulatory documentation, and the technical support hours you either get from your supplier or pay an engineer to cover. If it's not on your spreadsheet, you're not comparing vendors — you're comparing invoices.
Second: when a supplier tells you what they're not good at, write it down and trust them more, not less. That's not a weakness in the relationship. It's the only evidence you'll get that the rest of what they say has been checked.
Third, and this is the one I keep relearning: the conditions that made your last sourcing decision correct can quietly stop existing. Our surface prep changed. Our aluminum finish changed. Neither event triggered a review, because neither event looked like a procurement event. Build the review into the calendar anyway.
This all reflects our situation — a mid-size manufacturer with four product lines and fairly predictable demand. If you're running a job shop with constant substrate changes, or buying at a scale where you're dealing with producers directly instead of distributors, your math will look different from mine. Pricing and lead times here were accurate as of late 2024 and probably aren't anymore. Verify before you budget.
But the part about asking better questions? That one holds up.