How Not to Buy a Watch That Only Appears to Measure
A smartwatch can be a serious training instrument.
Or it can be a bracelet that displays invented figures with great conviction. The difference between the two is not visible in the shop, not visible in the specifications on the box, and most often not visible in the first weeks of use either.
This article is about how to tell them apart.
The Problem Nobody Mentions at the Point of Sale
The market has filled with cheap watches promising “over 100 sport modes.” It sounds generous. In reality, on most of them, all 100 modes do exactly the same thing: they start a stopwatch and read the pulse at the wrist. The rest is decoration.
One example makes it clear: swimming.
A cheap watch in “swim” mode will show you, at the end: 45 minutes, 320 calories. That is all.
A watch built for swimming shows you: the number of pool lengths, the stroke automatically identified (freestyle, breaststroke, backstroke, butterfly), the stroke count per length, the time per 100 metres, the rests between sets, and the SWOLF index — a measure of technical efficiency combining time with stroke count. That last one is the only figure that tells you anything about how you swim, rather than merely how much.
Both watches have an accelerometer. The difference is not the sensor. It is what the software knows how to do with its data.
The same logic applies everywhere: in running (cadence, vertical oscillation, ground contact time), in cycling (power, cadence), in strength work (set and rep detection), in skiing (number of descents, vertical drop). Every sport has its specific metrics, and those metrics are the reason you monitor at all.
Five Things to Check Before You Buy
1. Is the physical sensor there?
A watch without a barometer cannot measure floors climbed or elevation gain, however many sport modes appear in its menu — it will estimate them from GPS, badly. Without a magnetometer (compass) it has no true heading. Without a skin temperature sensor it cannot estimate nocturnal physiological variation.
The core sensors, in order of usefulness for someone who trains:
- Optical heart rate (PPG) — essential; every watch has one.
- Accelerometer and gyroscope — essential; cadence, movement detection and sleep analysis all come from these.
- Barometer/altimeter — needed for elevation, floors, sports over variable terrain.
- GNSS — for anything performed outdoors.
- SpO2 (pulse oximetry) — useful as a trend, not as a medical measurement.
- ECG — on higher-end models; see the caution below.
2. What quality is the sensor?
This is the invisible difference. Two optical heart rate sensors are not equivalent. What matters is the number of LEDs and photodiodes, the wavelengths used (green for pulse, red and infrared for oximetry), the sampling frequency and — most importantly — the algorithm that removes motion artefacts.
The practical test is simple: at steady effort, on a bicycle, almost any watch reports heart rate correctly. Under explosive effort with forearm contraction — which is exactly what happens in karate, kickboxing or bag work — weak watches lose the signal, smooth it artificially, or report half the true value.
If you train in impact sports and need accurate heart rate, the real solution is a chest strap, not a more expensive watch. The strap measures electrically rather than optically, and is an order of magnitude more accurate under variable effort. It costs little and pairs with almost any watch.
3. Positioning — GNSS, not GPRS
A point of terminology, since the two are often confused: GPRS is data transmission over the mobile network. GNSS is satellite positioning — the general term covering GPS (American), GLONASS (Russian), Galileo (European) and BeiDou (Chinese).
What matters today is not how many constellations are supported — nearly all watches support all of them — but whether reception is single-band or dual-frequency (L1+L5). Dual frequency substantially reduces errors in cities, between tall buildings, under tree cover and in valleys. If you run or cycle in town, it is the difference between an accurate track and one that shows you passing through buildings.
4. The software and the reports
The watch is the sensor. The value lies in what happens to the data afterwards.
Before buying, check what the app actually looks like:
- Are the reports useful or merely decorative? A colourful graph reading “sleep score 82” tells you nothing. A genuine breakdown by sleep stage, with a 30-day trend, tells you something.
- Does it analyse training load? The ratio between acute and chronic load is the best available instrument for avoiding overtraining.
- Does it sync with what you already use? Strava, Google Fit, Apple Health, nutrition apps.
- Can you export your data? This is the criterion nobody checks and everybody regrets. If you cannot extract your files (GPX or FIT format), you do not own your history — you rent it until the manufacturer shuts the service down or changes the subscription. Five years of training can vanish on a commercial decision you were never consulted about.
- Is there a subscription? Some advanced features sit behind a monthly payment. Not necessarily a bad thing, but you should know beforehand, not afterwards.
5. Certification, where it matters
The ECG function is where marketing is most aggressive and the difference greatest. A genuine ECG, on a device certified as a medical device in the European Union, can detect atrial fibrillation with clinically validated reliability. An “ECG” on an uncertified budget watch is an animation.
The same applies to oximetry: consumer SpO2 is useful as a trend over time, not as a spot value on which to base medical decisions.
The rule: if a very cheap watch advertises medical functions, ask where the certification is. If you cannot find it, it does not exist.
The Calorie Trap — and Why I Rewrote This Article
The first version of this text rested on the calorie expenditure recorded at two of my own training sessions. I have rewritten it, because I checked how reliable that figure is. It is not.
The reference study in the field (Shcherbina et al., Journal of Personalized Medicine, 2017), conducted at Stanford across seven devices and 60 subjects with indirect calorimetry as the standard: heart rate was measured accurately, with errors below 5% on most devices. Not one, however, achieved an error below 20% in estimating energy expenditure. Real errors ranged from 27% to 93%.
A more recent study on aerobic dance — precisely the kind of effort in our Cardio Bachata-Karate format — found errors between 16.6% and 42.8%, depending on the model.
The practical translation: the calorie figure on your watch may be a third higher or a third lower than reality. It is not a basis for a nutritional calculation, and not a basis for any decision.
What remains valid: comparison. If you measure two sessions with the same watch, worn the same way, the relative values are informative even when the absolute ones are not. In my case, a Karate–Bachata session came out at 6.55 kcal/min and a Tai Chi–Kizomba session at 5.4 kcal/min. The figures themselves say nothing. The ratio between them — roughly 20% higher intensity in the first — says something real, and that is exactly how a watch is worth using.
What Is Actually Worth Tracking
In order of reliability, from what is measured well to what is estimated badly:
Highly reliable: heart rate at steady effort, duration, GNSS distance, cadence.
Reliable as a trend: heart rate variability (HRV), resting heart rate, sleep architecture, training load.
Indicative only: estimated VO2 max, recommended recovery time, “stress scores.”
Unreliable: energy expenditure.
For anyone past forty, the most valuable figures are not the ones recorded during training but morning resting heart rate and HRV. Both change before any subjective sensation does. A resting heart rate five to seven beats above your own normal, several days running, means the body has not recovered — however well you feel. That is the signal that prevents injuries and illness, not the calorie count.
Heart Rate Zones — A Warning
Almost all watches calculate zones from the formula “220 minus your age.” It is a crude approximation, with a real deviation of ±10–12 beats in any given person. In someone trained for decades, it can be entirely wrong.
The consequence: if your watch consistently shows “zone 5, maximum effort” during a session you experience as moderate, do not push and do not panic — most likely the formula is wrong, not you. Real thresholds are determined by an exercise test at a sports medicine practice. It is worth doing once; you use the result for years.
Which Brand?
I give no models, because they change every year, and an article listing models becomes false faster than it is read. Brand characteristics, however, are stable:
- Garmin — the richest data ecosystem for sport, very good battery life, full data export. The app is dense at first.
- Polar — a long tradition in accurate heart rate measurement and exercise physiology, with an endurance focus.
- Apple Watch — the best integration with iPhone, certified health functions, a very broad app ecosystem. Battery life remains the weak point.
- Samsung Galaxy Watch — the Android equivalent, solid health functions, well integrated.
- Suunto — built for the outdoors and endurance, robust.
- Huawei — the best price-to-battery ratio, honest sensors. The software ecosystem is more closed, and data export more limited.
I currently use a Huawei Watch GT. It does what I need, with good quality sensors at a fair price — but I know what it does not do, and I take the calorie figure as an orientation rather than a measurement.
Three Questions Before You Pay
- Which sport do I actually want to monitor — and does this watch know its specific metrics, or does it merely start a stopwatch?
- Can I export my data in a standard format, if I change brand tomorrow?
- What is locked behind a subscription?
If the answers are good, price is secondary. A watch costing £150 that accurately measures what concerns you is worth more than one costing £500 that accurately measures something else.
Instead of a Conclusion
The watch does not train you. It shows you what you have done — and if it is a good one, it shows you what you failed to notice: that you have slept badly for ten days, that your resting heart rate has climbed, that you are training harder than you think and recovering less than you need.
That is what is worth having. Not the calorie count.
Do you monitor your training? If not, perhaps it is time to start — but begin with the question “what do I want to find out,” not “which watch should I buy.”