I handle purchasing for a 150-person contract manufacturing shop. My background isn't metrology; I'm a procurement person. That means I get asked to buy things like CMM machines, order Keyence sensors from a model number, and adjudicate the extremely unending debate between Mitutoyo vs Starrett calipers.
For most of my career, I treated these as standard orders. Get three quotes, compare lead times, ask what engineering really needs, pick one finance won't reject. But after a while I started noticing the same thing. The repeated failures in these purchases were rarely caused by the brand or the model. They were caused by the fact that I was treating a measurement process like an office supply order.
It Looked Like a Product Problem
The first big CMM machine request came from our QC supervisor. A customer audit was coming, and we needed traceable dimensional data. It looked simple: buy a machine. I researched axis travel, accuracy specs, software options, and sent out RFQs.
At the same time, an engineer asked me to find a Keyence manual so he could check whether a specific sensor would work on dark, glossy parts. He didn't need a brochure. He needed an application answer. And of course, we were also replacing calipers, which turned into a running argument about Mitutoyo vs Starrett calipers.
None of these requests are unusual. Each one is a purchase. But as I looked closer, none of them were really about the item itself. If we picked a different caliper brand, we still needed a calibration plan. If we bought the CMM machine, we still needed an operator and a measurement program. If we ordered the Keyence sensor, we still needed to test it on actual parts before trusting a production decision.
Our maintenance manager once forwarded me a thermal camera news today article about a manufacturer catching an overheated bus bar before failure. He asked if we should buy a thermal camera. Maybe. But without baseline images and a monthly scanning route, the camera is just an expensive flashlight.
The Real Problem Was Never the Tool
I'm not saying hardware is irrelevant. It is. A caliper with cheap electronics won't hold up in a shop with cutting fluid in the air. A CMM machine with too little travel can't measure a large part. A sensor with the wrong wavelength won't see what you need it to see.
The real issue on the buyer side is that we compare products before we compare processes. It's easier to ask 'Mitutoyo or Starrett?' than to answer 'How will someone prove to an auditor that the measurement was valid?' It's easier to request a quote for a CMM machine than to decide who will write the inspection program and what happens when a probe crashes. But the product decision comes after the process decision, not before.
Let's use calipers as an example. A brand comparison will show similar resolution, maximum measuring range, and price. What those spec sheets do not show: how long calibration takes, how easy replacement jaws are to get, and which brand the local calibration lab actually knows how to service. We learned this the expensive way after buying a popular model that required a six-week wait for a replacement blade.
With the CMM machine, the machine is only part of a measurement system. You need fixturing, probe calibration routines, temperature controls, operator training, and a report template your customer accepts. I almost bought one by comparing axis travel and price alone.
Honestly, I still don't understand why precision equipment quotes include so little of this. Maybe reps assume the buyer has already figured it out. In our case, nobody had.
What a Bad Purchase Actually Costs
A bad precision equipment purchase doesn't show up as a dramatic machine failure. It shows up as a process failure weeks later.
When I took over purchasing in 2021, we didn't have a formal process for checking calibration certificates. I ordered a calibrated instrument, but didn't specify what the certificate had to include. The certificate we received said PASS but didn't list the standard used or the measurement uncertainty. Our quality engineer rejected it. The customer's inspector agreed. The $2,400 instrument sat in its box while we waited for a replacement certificate, and our shipment was delayed.
The machine wasn't the problem. The paperwork was. That cost me more than money—it made me look like I didn't know what I was doing to my operations manager.
Then there is the people cost. A $50,000 CMM machine is not capacity if no one has been trained to program it. A Keyence sensor with the wrong setup can be a rejection machine. A thermal camera with good resolution but no emissivity training can tell you a hot panel is a cold panel.
Thermal camera news today is always full of 'look what this caught before failure' stories. What those stories leave out are the months of inspection discipline behind the image. The camera is the easiest part. The process is the actual investment.
What I Do Now
I'm not against spec sheets. I use them. But now I ask a different set of questions before contacting a supplier:
- What exactly are we trying to measure? Not just the part name—the dimension, condition, or trend that matters.
- How will the operator know the measuring tool is still accurate? The answer has to be a real calibration plan, not a sticker.
- If the reading seems wrong, what is the response? Who decides whether to stop the line, and what paperwork gets generated?
- For the CMM machine, what is included beyond the machine itself? I need probes, training, software, installation, and a sample run on our part.
- For Keyence sensors, I ask for an application demo or sample test. A Keyence manual tells me how to configure the sensor; it doesn't tell me if it will work on our unique surface and distance. I now let the sensor prove itself.
- For the Mitutoyo vs Starrett calipers debate, I compare service lead times and calibration turnaround before I compare prices.
This sounds simple, but when I started, none of it was done. My job changed from 'order what the engineer asked for' to 'understand what the engineer is actually solving.' Most of the time, they appreciate it.
I'd rather spend ten minutes asking questions than buy the wrong tool and deal with a failed audit three months later. An informed customer makes faster decisions. That's true when the customer is outside our company, and it's true when the customer is our own production manager.
Product is not process. The next time you compare brands or quote a machine, stop and ask how the measurement stays trustworthy after it leaves the shipping crate. If you don't have a good answer, no brand name changes that.
