When the numbers didn't add up
On a rainy Tuesday in January 2024, I sat in our conference room staring at a spreadsheet that didn't make sense. I manage the metrology and test equipment budget at a 140-person industrial components manufacturer—roughly $180,000 annually across CMM calibration, optical comparator upkeep, fiber optic sensor replacements, and various handheld instruments. Over the past six years, I've negotiated with more than 30 vendors and tracked every invoice in our procurement system.
2023's calibration spending had crept up 23% versus 2022. Not catastrophic, but not nothing. My boss wanted a forecast for Q2. Before I could build one, I needed to answer a more basic question: were we actually getting what we paid for?
The answer, as it turned out, was mostly yes—but with one gap wide enough to drive a delivery truck through.
The quote that looked reasonable (until it wasn't)
Our primary CMM—a workhorse we'd bought in 2019—needed its annual recalibration. The incumbent vendor, a local outfit we'd used for four years, quoted $4,200 for on-site service. That was up from $3,100 in 2021. I asked why. The account rep said "inflation and travel costs." Fair enough, I thought. Everything costs more now.
But something nagged at me. So I pulled quotes from two other providers:
- Local incumbent: $4,200 (on-site, 2-day turnaround)
- Regional lab: $5,800 (pickup/return, 5-day turnaround)
- Hexagon official channel: ~$6,500 (on-site, 3-day turnaround, full accreditation documentation)
The gap between the cheapest and most expensive was $2,300—not trivial for a single service event. I remember thinking: is the Hexagon premium worth 55% more than what we're already paying?
I dug into the incumbent's paperwork. That's when the first red flag appeared: their ISO/IEC 17025 accreditation scope didn't cover our specific CMM model's volumetric accuracy testing. They were subcontracting that part to a third party (i.e., the calibration report would carry someone else's stamp on the critical measurements).
I called them on it. The response: "That's standard practice." Maybe. But it meant our traceability chain had an extra link—one we weren't controlling.
The failure that forced a reckoning
Three weeks later, our quality manager forwarded an email from our largest customer. A batch of precision housings we'd shipped in February had failed incoming inspection at their facility. Their CMM readings showed dimensional drift outside our reported tolerances.
We spent the next 48 hours scrambling. Production halted on that line. Our quality team re-measured samples on our backup CMM (thankfully, we keep one calibrated independently). The parts were fine. The problem was in the calibration data we'd sent along.
The incumbent's report lacked a complete measurement uncertainty statement—the kind of documentation required by ISO 17025 for acceptance decisions. Without it, our customer's audit couldn't verify our results against NIST-traceable standards. They rejected the entire batch on paperwork grounds (ugh, again—this was the second time in two years).
We issued a corrective action, re-shipped with Hexagon-issued calibration certificates (we'd rushed a calibration through their official channel as a stopgap), and ate $8,400 in expedited freight, re-inspection fees, and administrative overhead.
That $2,300 we "saved" on the cheaper quote? It cost us almost four times that in real money—plus a relationship hit I'm still repairing.
"If I could redo that decision, I'd have paid for the accredited calibration upfront. But given what I knew then—we'd used the incumbent for years without a major issue—the choice felt low-risk. That's the trap: familiarity masquerading as reliability."
What I changed (and what I'm still figuring out)
I have mixed feelings about how we handled the vendor transition. On one hand, switching to Hexagon's calibration service was clearly the right call for our primary CMM—their documentation meets the audit requirements our customers actually check, and the three-day turnaround means less production downtime. On the other hand, I should have caught the accreditation gap during our 2022 renewal, not after a customer rejection.
The risk calculation I ran back in January was flawed. I kept asking myself: is $2,300 worth potentially disrupting a four-year vendor relationship? But I framed the question wrong. It should have been: what's the probability that non-accredited calibration data fails a customer audit, multiplied by the cost of that failure?
Even a 10% chance of an $8,400 problem has an expected cost of $840—which, added to the $4,200 sticker price, makes the $6,500 accredited option look like a bargain.
We now require three things from any calibration vendor:
- ISO/IEC 17025 accreditation scope that explicitly covers our instrument category (I verify this on the issuing lab's official website, not just the vendor's marketing materials).
- Complete measurement uncertainty statements for every reported value.
- Written confirmation of whether any measurement is subcontracted—and if so, to whom.
This hasn't just applied to CMMs. When we audited our optical comparator service last quarter, I found similar gaps in two of the three quotes we received. The cheapest option had no uncertainty documentation at all. We went with the mid-priced lab that did.
Fiber optic sensor replacements were less dramatic, mostly because those are component swaps rather than calibration services—the specs are what they are. But the same principle applies: if you can't verify the traceability, you're trusting a black box.
The durability question I still can't fully answer
One area where I'm somewhat stuck: comparing the durability of industrial photoelectric sensors across brands. We've used ifm sensors in a few of our automated test fixtures for years, and they've held up fairly well. But when I tried to evaluate whether they actually last longer than comparable units from other manufacturers, I hit a wall.
Publicly available data on real-world sensor lifespan is... thin. Most manufacturers publish MTBF figures derived from lab conditions (controlled temperature, no vibration, clean power). Our floor is not a lab. We have coolant mist, vibration from nearby presses, and voltage sag during shift changes.
I asked three vendors for field failure rate data. Two said it was "proprietary." One gave me a presentation slide with no sample size or methodology. So my current approach is anecdotal: we track replacements in our maintenance log and compare after 18-24 months. Not scientific, but it's what we have.
If anyone has a better method for evaluating sensor durability in harsh environments, I'm genuinely interested. This is one place where I think our procurement process still has a blind spot.
What this whole mess taught me
Looking back, the calibration incident wasn't really about one vendor being bad or another being perfect. It was about asking the right questions before signing the PO, not after something breaks.
If you're managing a metrology budget, here's the test I now apply: if the calibration report can't survive a customer audit without extra explanation from me, it's not actually calibrated for my use case. Period.
The cheapest quote usually isn't the lowest total cost. That sounds obvious written out like this. But when you're staring at a spreadsheet and everything looks equivalent on paper, the gap between $4,200 and $6,500 feels like a lot of money. It took an $8,400 mistake to remind me that it isn't.
