Application guidance for water, heat and gas monitoring programs with audit-ready evidence.
Itron application note cover
Application Note

Why Your Industrial Sensor and Meter Purchases Go Wrong — And It's Not About the Price

2026-11-03 by Sofia Lindberg

Last fall, our maintenance lead handed me a list: twelve Balluff fork sensors for the packaging line, two 87V true rms multimeters for the electrical team, and "some more IO-Link cable" (his exact words). It looked like a routine order. Three weeks later, I was dealing with two RMAs, a half-shift production halt, and a question from finance I couldn't answer.

I've been doing this job for six years. I manage all the ordering for this plant — roughly $180K a year across a dozen suppliers — and I report to both operations and finance. Before last fall, I thought I was good at it. After last fall, I realized I'd been doing a different job than the one I thought.

I thought I was doing spec matching. In reality, I was doing "reading."

The surface problem: it looks like a spec-matching job

From the outside, buying industrial sensors and meters looks straightforward. An engineer gives you a model number. You check a spec sheet. You compare two or three vendors on price and lead time. You place the order. The parts show up. The engineer installs them. Everyone moves on.

The whole process rests on an assumption: that what's on the spec sheet is what the plant floor needs. I believed that assumption for five years.

The 18% price increase and the three extra hours of downtime made me stop believing it.

The deeper problem: a spec sheet is a lab report, not a plant floor

Here's the thing nobody in procurement explains to you: sensor specs are tested under controlled conditions. 25°C. Clean air. Stable voltage. Standard mounting distance. A plant floor is none of those things. A plant floor is 50°C inside a cabinet next to a welder, with a VFD humming ten feet away and vibration coming through the frame.

Three specific versions of this problem showed up on my desk.

1. The Balluff fork sensor that "fit" but didn't work

The spec sheet for a Balluff fork sensor lists detection distance, slot width, response time, and light source type. Those are the things you can compare. What it doesn't list is what's actually inside the slot: a clear label on a clear backing, a mirror-finish component, or a weld flash that occasionally crosses the beam. None of that is on the spec sheet, and all of it decides whether the sensor works.

I bought twelve of them (correct model, correct spec). Six kept triggering on reflections. Not the sensor's fault. Not the engineer's fault. My fault, for treating "matches the requested model number" as the same thing as "suitable for the installation."

2. "Balluff IO-Link" is a protocol, not a plug

This one still bothers me. The Balluff IO-Link sensors we bought last year were the right part number, IO-Link 1.1 compliant, and — on paper — fully compatible with the controller. When they arrived, our controls tech put them on the shelf for a week because the IO-Link master module hadn't been installed on the controller yet.

The sensor was an IO-Link device. Fine. The problem was that "IO-Link device" doesn't mean "plug it in and go." It means the device speaks IO-Link to a master, and the master isn't free. It has to be on the controller, it has to be wired, and the port has to be configured. Three separate purchases, not one.

If you're a buyer, "compatible" is a word that needs three follow-up questions, not zero.

3. Reading is a skill, and nobody trains procurement for it

This ties back to something I get asked a lot outside of work: how do you read a Sensus water meter? On the surface it's a completely different topic. In practice it's the same problem. The dials are just numbers. If you don't know which dial is units and which is tens of units — and if you don't know what "reading around the dial" actually means — you write down a number that's wrong, and every decision downstream of that number is wrong.

A colleague of mine handles the utility billing for one of our satellite locations, and she ran into exactly this last quarter. The meter was fine. The reading wasn't. Three months of overcharges before anyone caught it.

Same story with a Coriolis flow meter. Same story with an 87V true rms multimeter. The reading is not the measurement. The reading is the interpretation of the measurement. A Coriolis meter installed without a zero check is a number with a lot of confidence and no accuracy. A standard multimeter on a VFD output reads "average" voltage on a waveform that isn't sinusoidal, which means the number your tech writes down is not the number that's actually there.

That's why the 87V true rms multimeter showed up on the list in the first place — not because it's a Fluke, but because the electrical team is measuring drive output, and a cheap meter will lie to them in a way they can't see. Standard 87V units run $230-290 from major industrial distributors (2024 Q4 quotes; verify before ordering).

What "wrong readings" actually cost

Here's the short version: the cost of a wrong reading does not show up on the PO.

  • Phantom sensor triggers. A fork sensor tuned to the wrong slot configuration doesn't fail — it just triggers at the wrong times. You don't lose a part, you lose a shift to debugging. Ours cost us about half a day on the packaging line (this was October 2024, if I remember correctly).
  • Coriolis drift. A Coriolis flow meter without a proper zero check reads, say, 1.2% high on day one. That's within tolerance. By month three, it's 4% high and the invoices it's feeding are wrong. Nobody notices until a customer complains.
  • Multimeter misreads. A non-true-RMS meter on a distorted waveform reads high or low depending on the harmonics. Your tech writes down "24V, 60Hz." What's actually there is something else. The diagnostic decision that follows is built on a bad number.
  • Utility bill errors. A miscounted Sensus water meter — the kind of thing that happens every month at a facility — turns into a three-month dispute with the utility. Nobody catches it until someone reads the fine print on a bill.

None of these are "wrong product" failures. All of them are "right product, wrong reading" failures. And in my experience, the second category is bigger — and harder to see — than the first.

The most frustrating part of this: the same things kept happening across completely different product categories. A sensor that seems fine on the line. A meter that "reads one way" when it was reading another. You'd think the fix would be a stricter spec review. It wasn't. The fix was understanding that reading is a separate step from buying.

What I do differently now

I don't have a ten-point framework. I have three questions I ask before any sensor or meter order goes out.

  1. What does this thing actually live in? Temperature range, vibration, ambient light, media, and interference source. Not the spec sheet's "operating temperature" — the actual cabinet temperature on the line. If the answer is "standard," I ask again. It's usually not standard.
  2. What does it talk to, and is that thing already installed? For anything IO-Link, this means the master, the cable, the port, and the configuration tool. "Compatible" is a sales word. Installed, wired, and configured is the operational word.
  3. Who reads this number, and how will they know when it's wrong? This is the 87V true rms multimeter question, the Coriolis flow meter question, and the Sensus water meter question, all in one. If nobody can tell whether the reading is wrong, you haven't bought a measurement. You've bought a guess with a digital display.

Two of those three questions aren't answerable from a spec sheet. The third is only answerable by the person who's going to install the thing. That means the real work of sensor and meter procurement isn't matching part numbers — it's pulling the right people into the conversation before the order goes out.

That sounds slower than just matching model numbers and hitting submit. In practice, it isn't. The half-shift of downtime we spent debugging phantom triggers was the slower version — it just wasn't on my calendar where I could see it coming.

Total cost of ownership on measurement equipment isn't just the unit price, the installation, and the calibration. It's also the cost of every decision made on top of a reading that turned out to be wrong. Once you start counting that, the cheap option gets a lot more expensive.

Prices cited are from 2024 Q4 distributor quotes; verify current pricing and lead times before ordering.

Sofia Lindberg

Sofia Lindberg

Sofia Lindberg is a thermal and environmental measurement analyst covering thermal imaging cameras, infrared thermometers, hygrometers, moisture meters, weather stations, and air quality meters. She uses ISO 18434-1 thermography principles and evaluates emissivity control, thermal sensitivity, spatial resolution, field of view, response time, and sensor drift. She guides reliability, building-diagnostics, and EHS teams in selecting instruments and interpreting readings without overstating what a single measurement can prove.

Leave a Reply