A number on a screen feels like the truth. It isn't — not until you've earned the right to believe it. A reading you can't trust isn't data; it's decoration. And a sensor that lies is worse than no sensor at all: an empty hook keeps you watchful, while a confident wrong number sends you to bed easy on the night the heater fails.
Two ways to be right
A gauge can be right — or wrong — in two different ways, and the difference decides what you're allowed to ask of it.
Accuracy is how close the reading is to the real thing: a thermometer that says 70 when it truly is 70. Precision is how consistent it is: a thermometer that reads 73 every single time it's truly 70 is rock-steady — and wrong by the same three degrees every time.
Picture a dartboard. Accurate and precise is a tight cluster on the bullseye. Precise but not accurate is a tight cluster off in the corner — and that's the dangerous gauge, because it looks dead reliable while it lies the same amount all day long. Accurate but not precise is scattered loosely around the center: noisy, but honest on average.
Which one you need depends on the job. “Is it warming up?” only needs precision — even a sensor three degrees off will show the climb. “Mix this feed to exactly this strength” needs accuracy — consistently wrong ruins the batch. Know which kind of right your decision is asking for before you trust the number.
Sensors drift — the quiet kind of wrong
A sensor that was right in January can be lying by June and still look perfectly fine. That's drift, and it's the failure that bites hardest, because the number stays believable the whole way down. Nothing flashes red; the reading just slowly turns into fiction.
Some gauges drift faster than others. A pH probe is the worst offender — it can wander in a couple of weeks of use, and the glass ages on the shelf whether you touch it or not. Cheap humidity sensors creep as they age and as dust settles on them. CO₂ and other gas sensors wear out. So calibration isn't a thing you do once when the box arrives. It's a habit, on a schedule, for as long as you lean on the reading.
The cheap-sensor truth
A $12 sensor isn't a bad sensor. It's an honest one — if you know its job. Appropriate technology cuts both ways here: that cheap sensor will faithfully show you a trend — the greenhouse is warming, the humidity is climbing toward mold weather — and a trend you can trust is worth far more than a precise number you check once a year.
So match the gauge to the decision. For “is something going wrong?” a cheap sensor and a clear trend is plenty, and spending more would be gold-plating. For “dose this exactly” or “prove this number to a buyer,” you need a calibrated, accurate instrument — and that's where the money is well spent. Don't gold-plate the easy calls; don't cheap out on the exacting ones.
Earning a reading's trust
You don't have to take a gauge's word for it. Five plain habits turn a number into something you can act on.
Calibrate against something known. You don't need a lab. For pH, two buffer solutions (a 4 and a 7) tell the probe the truth. For humidity, the salt test: seal a little damp table salt in a jar with the sensor and the air inside settles at 75% — a sensor reading 60 in that jar is fifteen points low until you correct it. For temperature, ice water is 32°F (0°C) and a rolling boil is 212°F (100°C) near sea level. Cheap references, real truth.
Recalibrate on a schedule, because drift is real. A working pH probe wants checking every week or two; slower gauges, once a season. And write down when you last calibrated and against what — that date is part of the reading. Data is king only if the king is trustworthy, and the calibration log is how you know.
Cross-check — two cheap eyes beat one you never blink. Two $12 sensors that agree are more trustworthy than one expensive one you never verify. When two disagree, you've learned something instead of quietly trusting a lie. A second sensor is almost always cheaper than being wrong about a crop.
Set sanity bounds. Know what's impossible. A pH of 14, a humidity of 130%, a greenhouse at 200°F — those aren't the world, they're the gauge. Decide the plausible range up front and don't act on anything outside it.
Learn the tells of a dying gauge. A value that freezes (the sensor stopped sensing but keeps reporting its last guess). A reading pinned hard at 0 or 100. Numbers that suddenly get jumpy, or lag far behind reality. A gauge that no longer agrees with its cross-check. Any one of these means check it before you believe it.
When you can't fully trust it, change the job — don't toss it
A gauge you only half-trust isn't garbage. Demote what you ask of it. Let a drifting sensor answer “is the trend heading the wrong way?” — an alert — and keep it away from “hold this exact setpoint,” which is control. Be honest about each gauge's grade, and let the trustworthy ones carry the decisions that actually matter.
Quality engineers have a whole discipline for this; they call it measurement-system analysis. We just call it trusting your gauge — and it's the cheapest credibility a grower can buy. It's also the first real move in the loop behind everything here: measure, but trust the gauge first.
Common questions
Are cheap sensors accurate enough for growing?
Often, yes — for the right job. A $12 sensor reliably shows a trend, so for "is something going wrong?" it is plenty. For exact dosing, or proving a number to a buyer, calibrate it against a known reference or step up to an accurate instrument. Match the gauge to the decision.
How often should I calibrate a pH sensor?
In active use, about every week or two — pH probes drift quickly and the glass ages even when unused. Calibrate against two buffer solutions (a 4 and a 7), and write down the date so you know how fresh the reading is.
Why do two of my sensors read differently?
Usually one or both has drifted, or they simply have different accuracy. The disagreement is useful: it tells you not to trust either one blindly. Cross-check both against a known reference, then correct or retire the one that is off.
What is the difference between accuracy and precision in a sensor?
Accuracy is how close a reading is to the true value. Precision is how consistent it is. A sensor can be precise but wrong — off by the same amount every time — which is the most dangerous kind, because it looks reliable. Spotting trends needs precision; hitting exact targets needs accuracy.
How do I calibrate a humidity sensor without special equipment?
Use the salt test: put a little damp table salt in a sealed jar with the sensor and wait a few hours. The air inside settles at 75% humidity — so whatever the sensor reads, the gap from 75 is its error, and you can correct for it.
How do I know if a sensor is broken or just reading wrong?
Watch for the tells: a value that freezes, a reading pinned at its minimum or maximum, sudden jumpiness, a slow response, or disagreement with a second sensor or a known reference. Any of these means check it before you act on it.