Why I Check Every Kennametal Order Before It Touches a Machine

If you ask me, the most dangerous part of a machining operation isn't the spindle. It's the assumption that the tool in your hand is the tool you ordered. I'd argue that the worst quality problems in metalworking are born at the desk, not at the machine. And that's why I check every Kennametal order before it touches a machine.

I'm a quality/compliance manager at a job shop that machines parts for the energy and aerospace industries. We run about 40,000 parts a year. Roughly 200 tooling order lines cross my desk annually, and in 2024 I rejected 11% of the first deliveries we accepted into inventory. They didn't all fail on dimensions. Some had the wrong coating grade. Some had the wrong geometry. One had a chip on the edge that was hidden under a loose wrapper. The machine would have found all of them eventually. That's the problem.

The expensive part starts before the spindle

There's a boring, unglamorous layer of quality control that happens before a CNC program even loads. It involves reading packing slips, checking batch numbers, and comparing the insert grade on the box to the routing sheet. It is not exciting. But honestly, it's the layer that stops the money from leaking out.

Let me give you a real example. Last year, I assumed 'same as previous order' for a carbide turning insert. It was the same Kennametal KC5010 coating and the same insert material. I didn't verify the chipbreaker geometry. Turned out the new order was a different insert shape. The wrong insert made it to the machine, touched a part, and left a finish that was out of tolerance. The part scrap. The tool was 'fine' if you only looked at wear. But the geometry was wrong for the pass.

I still kick myself for that one. If I'd spent five minutes with a digital caliper and the technical drawing, we would have caught it before any steel came off. Instead, we lost a part, had to re-prove setup, and filed a rework claim that took three weeks to resolve. I know the number: that mistake cost us about $8,000 in material and labor. It also cost me the ability to say 'that never happens here.'

What a receiving check actually looks like

We didn't always have a formal receiving check for tooling. The third time we got a wrong order—not from Kennametal specifically, but from a distributor—I finally created a verification checklist. It's not complicated. It's just the 12-point list that should have existed from the beginning:

• Purchase order number matches the packing slip
• Insert grade printed on the box matches the approved grade on the routing
• Coating code matches the spec and the application
• Geometry and chipbreaker match the drawing
• Edge prep and corner radius are within stated limits
• Batch code can be traced back to the material cert

That's the list. No microscope, no CMM, no lab. It takes 10 minutes for a standard order. And it's caught more issues than I can count.

Kennametal Henderson and the email habit

When a delivery arrives from the Kennametal Henderson facility—that's the plant in Henderson, North Carolina—the paperwork is usually consistent. But 'usually' is not a spec. I still check. One shipment from Kennametal Henderson had a subtle mismatch between the coating code printed on the box and the coating code on the insert itself. The product was fine; the labeling was off. If we'd skipped the check, the wrong traceability record would have linked a good insert to the wrong process.

That's why I tell people to keep the Kennametal email address for technical support saved in your contacts. I know it sounds like a minor thing. But when you need to ask 'Can I use this grade for an interrupted cut?' you want a written answer you can attach to the job folder. I emailed Kennametal once about a drilling grade for a 316 stainless job. The application engineer replied with a recommendation and a note about coolant pressure. That email became the reference for the whole production run.

Email is also better than a phone call for quality people because it leaves a trail. If a tool fails and the machine's program was correct, you want the evidence. A 'trust me' phone call doesn't hold up in a corrective action report.

Experience is great. Process is better.

We have a machinist named Henry. He's been running metal since before I was born. He can read surface finish by dragging a thumbnail over the part. Nobody argues with Henry on the floor. But Henry's age and experience don't make him infallible. Last month he signed a delivery receipt without opening the box. His reasoning: 'It's from a supplier we use all the time.'

To be fair, it was from a supplier we use all the time. But the box was wrong. Inside was an insert edge prep that didn't match the job. We caught it because a second person—someone with no seniority—followed the checklist. The fix wasn't telling Henry to trust the checklist. The fix was making the checklist a mandatory step, not an optional act of respect.

This is where I always think of the 'Lewis vs Waldo' argument. Two of our veteran machinists, Lewis and Waldo, used to disagree about how much verification was needed before a run. Lewis said 'I can feel it.' Waldo said 'measure it.' They were both good at their jobs. But when a disagreement is based on personal instinct, you get inconsistent results. The 'Lewis vs Waldo' debate didn't end when one of them won. It ended when we wrote the acceptance criteria into the work instruction. Now no one has to argue.

I'm not saying people can't be trusted. I'm saying a quality plan based on memory is a plan with a built-in failure date.

What about the 'we can't check everything' objection?

I hear this a lot: 'If you inspect every tool order, you'll slow down the shop.' Granted, there is a cost to checking. But you don't check everything. You check the critical 20% of variables that cause 80% of the problems.

We have a Monarch lathe in the shop from the 1960s. It's old, heavy, and still accurate. But we don't trust the Monarch because of its age. We trust it because we run a controlled test part before every critical job. The test takes twenty minutes. The last time we skipped it to meet a deadline, the part failed and the deadline got worse. That's not a machine tool story. That's a prevention story.

Roughly speaking, 80% of the tooling problems I've logged in four years happened in ordering, receiving, or setup—not in cutting. I don't have a peer-reviewed study for that. Take it with a grain of salt. But if that ratio is even close to true, the highest-ROI quality activity isn't a better spindle. It's a better question at the receiving dock.

Prevention is cheaper. It's also less heroic.

There's something satisfying about catching a batch before it reaches the machine. But it's quiet. No alarms. No rework heroics. Just a box set aside and an email sent upstream. That kind of work doesn't get the same attention as the night shift saving a crashed fixture. But it's real.

I'm not here to tell you Kennametal is the only tool manufacturer worth considering. I'm not here to attack anyone else, either. What I'm saying is that the tools are only as reliable as the system that checks them. Kennametal's technical data is detailed enough, and their application support is responsive enough, that a good verification system becomes possible. You still have to build the system.

So here's my position, restated: The check is not the slow part. The rework is the slow part. Five minutes of verification beats five days of correction. If you've ever had a tooling mistake cost you a part and a week, you know what I mean. And if you haven't—keep checking. That's how you stay lucky.