Why I Check Everything Twice: The Real Cost of Skipping Verification

I'd rather reject ten good units than accept one bad one. That's not a slogan—it's the arithmetic of rework.

I'm a quality and compliance manager at a water-quality instrumentation company. I review every deliverable before it reaches customers, roughly 200 unique items a year. I've rejected 7% of first deliveries in 2025 so far due to specs that were visibly off. Normal tolerance in most cases is tighter than the vendor wanted to believe. The vendor said "within industry standard." I said, "Not within ours." They redid it at their cost.

That experience shaped my opinion: the most expensive words in any operation are "she'll be fine." A five-minute verification beats a five-day correction. I don't think this is controversial, yet I keep seeing organizations skip the check because they're "too busy." I've been there. But I've also been the one paying for the aftermath.

So let me argue for prevention over cure—not as a slogan, but as a practical operating principle. I'll use three instruments that show up in my world: YSI sensors, an absolute encoder AFM60A, and a Mitutoyo micrometer. They're different devices, but the lesson is the same.

YSI Sensors: Calibration Is Not an Event

I work with a lot of YSI equipment. The YSI multiparameter sondes in our catalog measure pH, dissolved oxygen, turbidity, nitrate, and algae in the field. The optics are genuinely good. I've seen a YSI EXO sonde hold stable data for weeks in challenging conditions. But "stable" is not the same as "accurate forever."

Everything I'd read about advanced optical sensors said they require less maintenance than the old electrochemical membranes. That's true to some extent. In practice, though, you still need a real calibration routine before deployment. We check zero and slope against known standards. We log who did it, when, and with which lot of standards. That record matters more than the number itself.

If you're using a YSI sonde for compliance data, the calibration log is part of the data. A reading without a calibration trace is just a number. I've seen a project delay by two weeks because someone couldn't prove when the probe was last calibrated. The data itself looked fine. The metadata was missing. That's when I stopped treating calibration as a "maintenance task" and started treating it as a data requirement.

Also, keep the software current. I know it's tempting to ignore the "new version available" pop-up, but the YSI download portal exists for a reason. The firmware and software updates often fix edge cases that only show up in certain data-logging scenarios. I learned that the hard way: a logger reported pH values that shifted after a storm event. It wasn't the sensor. It was a firmware bug that had been fixed in the latest release. If I'd downloaded the update before deployment, we'd have avoided a lot of re-processing.

The Absolute Encoder AFM60A Remembers—But Only If You Check

A colleague in the process-control side asked me a question that stuck with me. He said, "If I use an absolute encoder, I don't need a home switch anymore, right?"

I get the logic. An absolute encoder AFM60A retains position information after a power loss. That's its whole purpose. But "absolute" refers to position encoding, not immunity to physical problems. The coupling can slip. The shaft can loosen. The mounting bracket can shift. The battery that maintains the multiturn count can die.

The conventional wisdom is that absolute encoders reduce commissioning time because you skip the homing routine. My experience says otherwise, at least when the encoder is part of a safety-relevant or quality-relevant motion system. You still need a start-up verification routine. It doesn't have to be a full home switch, but it should be a known reference check. If you're using an encoder IO Link connection, you get diagnostic data—signal quality, temperature, operating hours. That's great. But someone has to read it and act on it.

The most useful moment isn't when the encoder fails. It's the first time you see a warning flag. If you don't have a baseline from the initial commissioning, that warning flag is just a blinking light. A checklist from day one gives you something to compare against.

How to Calibrate a Mitutoyo Micrometer: Precision Is Not Accuracy

Now the tool everyone thinks is simple. A micrometer is a precision instrument. But precision is not accuracy. A micrometer can repeat the same wrong measurement all day if it isn't calibrated.

I've heard the question "how to calibrate Mitutoyo micrometer" many times. Usually the person expects a quick trick: close the spindle, check the zero, done. That's the start, not the end.

This is what I do—and I should note, I'm not an official Mitutoyo service center, so verify with your own metrology procedures and applicable standards. For compliance work, your calibration records should also be traceable to a recognized standard, such as NIST in the U.S. or the applicable ISO/IEC requirements. Verify the current version of the standard before you report data.

• Clean the spindle faces and the anvil. A fingerprint changes the reading by more than people want to admit.
• Check the zero on a certified gage block, not just on the closed spindle. Use a standard close to the size you actually measure.
• Take multiple readings across the measuring range, including a mid-range standard and a larger standard. A micrometer can have wear patterns that don't show at zero.
• Record the temperature. Temperature is the silent variable in precision measurement. A micrometer that's been in a hot toolbox reads differently than one left on a cold bench.
• Document everything. If there's no record, the calibration didn't happen.

That last bullet matters more than the procedure itself. I've seen a field engineer swear his micrometer was fine because it "clicked right." But the gage block set was overdue for recertification, and every measurement was traceable to a standard that had drifted. That's not a tool failure; it's a process failure.

The good news: you don't need to do this every week. The question is frequency. A weekly zero check and an annual dimensional calibration is common for moderate use, but verify with your quality system and OEM guidance. The point is that "how to calibrate Mitutoyo micrometer" doesn't have a single universal answer. It depends on your tolerance, your environment, and your risk.

But We Don't Have Time

I hear the objection coming: "We don't have time to check every sensor, encoder, and micrometer." I've felt that pressure too.

Here's the counterintuitive part. The time spent on verification is rarely the bottleneck. The bottleneck is usually the rework that follows a missed check. When we implemented a 12-point pre-ship verification checklist in 2022, I was worried it would slow us down. It didn't. It did the opposite. The first quarter was nerve-wracking. I kept second-guessing: what if we were checking the wrong things? I didn't relax until I saw the numbers. The rate of field-reported issues dropped by a measurable amount. Not a dramatic change, but a real one.

I want to say the estimated saving was around $8,000 in potential rework in the first year, but don't quote me on that exact number—I'd have to pull the report. The important part is the direction: catching issues earlier costs less than catching them later.

My Checklist Isn't Perfect, But Memory Is Worse

I'm not saying every device needs a 12-point check before every shift. That would be ridiculous. But I am saying that if a process is worth doing, it's worth defining. A checklist is the least expensive insurance you can buy. It doesn't guarantee you'll never make a mistake. It guarantees you won't make the same mistake twice because you relied on memory.

Yes, there's a cost: time, paperwork, and the occasional feeling that you're being paranoid. But I've seen the alternative cost a lot more. In 2024, a batch of data loggers went out with the internal clock set wrong by six minutes. The sensor data was perfect. The timestamps were unusable. The customer had collected a week of data before anyone noticed. The clock check was already on our list—we still caught it only because we had a list. If we hadn't, we wouldn't have caught it at all.

So here's my opinion, stated as clearly as I can: prevention beats cure, every time. The five minutes you spend checking a YSI sensor before deployment, confirming the encoder IO Link diagnostics, or zeroing a Mitutoyo micrometer is not wasted time. It's an investment in not being the person who explains why a problem wasn't caught earlier.

Check the instrument. Check the log. Check the standard. And when you're done, check once more. It's boring. It's unglamorous. It's also how good companies stay good.