The Quote Was Clean. Then the First Quarter Closed.
Two probes. Same stated accuracy. A 38% price spread. We bought the cheaper one—not to save money, honestly, but because the spec sheets read like they came off the same template. (Should mention: "same template" was my reading. The vendor never said the parts were equivalent.) By the end of Q1 we were recalibrating every three weeks instead of every three months. Nobody had quoted us that line.
That was 2023. I've been less polite about procurement documents ever since.
The Surface Problem: Every Comparison Uses the Wrong Column
When a team sits down to choose a water quality sensor, or a centrifuge, or anything else on the bench, the conversation lands in the same place: accuracy, range, IP rating, price. Four numbers, side by side, one decision.
Accuracy, it turns out, is a snapshot. Stability is the season.
A YSI sensor that holds calibration through a full deployment cycle and a cheaper sensor that holds it for six weeks can carry the same printed accuracy specification. Both numbers are technically true. Only one survives contact with a river.
The Deeper Cause: A Spec Sheet Describes a Moment
Here's the assumption that trips people up: accurate instruments cost more because they're built to higher precision. That's backwards, or at least incomplete. Instruments that stay accurate—through biofouling, temperature swings, long deployments, and whatever gets sucked through the sensor head—are expensive to engineer. Precision at the factory is common. Precision at month eight is rare.
This is where the technology actually shows up. Optical platforms like the YSI EXO NitraLED UV-LED nitrate sensor were built around that exact problem—no reagents, no lamp replacement cycle, and a drift profile tight enough to trust over a field season rather than a single grab sample. That's not a spec sheet line. That's the difference between a year of data and a year of re-dos.
The same principle runs in the lab. A pipette that delivers the stated volume on day one is table stakes. One that still delivers it after two years of daily use, correct cleaning, and proper calibration is a different product.
Where It Actually Costs You
Three lines, in order of how badly they bite.
Calibration. Every sensor drifts. Every pipette drifts. Every reference thermometer drifts. The question isn't whether, it's how fast—and whether you find out before the data leaves the building. A dual-channel 52-series thermometer used as a bench reference is only a reference if it's traceable and in date. Otherwise you're measuring your own confidence.
Cleaning. This is the one people underestimate, because it looks like housekeeping. It isn't. Ask how to clean an Eppendorf pipette properly and you'll get an actual procedure: partial disassembly, specific cleaning agents, and seals that don't survive an autoclave. Skip a step and the pipette still "works"—it just delivers a slightly different volume. That error doesn't show up in the cleaning log. It shows up in the assay three weeks later, and by then you're re-running a batch and explaining why the numbers moved.
The same logic applies to a 5810 centrifuge. Service intervals are in the manual, not in the purchase order. Rotor inspection, seal replacement, imbalance history—all of it is maintenance, and none of it appears on the quote you compare bids against.
Downtime. A sensor out for service is a hole in your monitoring record. A centrifuge down is a stalled batch. A pipette out for calibration is a tech logging numbers by hand and a reviewer asking why the file looks different. The invoice for the instrument is a one-time event. The invoice for the downtime arrives every quarter.
The Arithmetic Nobody Quotes
When we reviewed our 2024 service contracts, the pattern was consistent: first-year cost of ownership ran roughly the purchase price plus 8–12%. Year two was higher. Year three was higher again—mostly calibration, partly consumables, and a meaningful chunk in expedite fees we paid because we hadn't planned ahead. None of that appeared in the original bid comparison. All of it was predictable if anyone had asked.
I should be careful here. I'm not a metrologist—I can't tell you the exact drift curve of a given electrochemical cell, and I'd push back on anyone who claims a single number applies across applications. What I can tell you, from the QA side, is that the variation between vendors on stability is far larger than the variation on accuracy, and almost nobody asks about it. Buyers compare the number that's printed. They don't compare the number that's earned.
What to Ask Before the Price
If you remember nothing else from this: ask what's not included before you ask what's included.
A short list that has saved us real money:
- What does calibration cost in year two and year three—not year one?
- What's the service response time, in writing, and what's the expedite fee when it slips?
- Which parts are consumable, and what's the annual cost of keeping them in stock?
- What's the cleaning procedure, and does it require anything we don't already own?
- What's the documented drift specification over a full deployment or service interval?
None of these questions makes a vendor look bad. The answer tells you which vendor has actually thought about it—and which one is quoting you a price that only covers the fun part.
The Part That Sounds Like a Sales Pitch (It Isn't)
The highest total quotation we've ever received turned out to be the cheapest contract we've ever signed. It wasn't close. The upfront number was 22% higher than the next bid; the three-year number was 14% lower, and the data quality was noticeably better because we weren't spending half the field season chasing drift instead of collecting readings.
That's not a pitch for paying more. It's a pitch for making the vendor put the whole number on the table. A price that only covers year one isn't a price—it's a down payment with a surprise attached.
Ask for the three-year number. Ask what's not included. The vendor who answers without flinching is the one you want. Done.