What Fitness Trackers Actually Measure and How Accurate They Are

Fitness trackers have moved from curiosity to mainstream hardware. Most people who wear one check it regularly and use the numbers to make decisions about how much they move, sleep or exercise. The problem is that the confidence the display projects does not always match the confidence the sensor deserves. Understanding what is actually being measured, and how, makes the data considerably more useful.

The core sensors in a modern fitness tracker are a photoplethysmography (PPG) sensor for heart rate, an accelerometer for movement detection, and in most current devices a gyroscope as well. Higher-end devices add GPS, skin temperature sensors, blood oxygen estimation (SpO2), and electrocardiogram (ECG) capability. The base sensors are the same across most of the market; the differences are in how each manufacturer processes the raw data and presents the result.

How Optical Heart Rate Sensors Work

The PPG sensor works by shining green light into the skin and measuring how much is reflected back. Blood absorbs green light more than surrounding tissue, so as the heart pumps blood through the capillaries under the skin, the reflected light fluctuates at the same frequency as the heartbeat. The sensor converts that fluctuation into a beats-per-minute reading.

This works reasonably well at rest. During light to moderate activity, the accuracy degrades. The problem is motion artefact: any movement of the wrist relative to the sensor introduces noise into the optical signal. The tracker's software has to separate genuine cardiac signal from the noise introduced by arm swings, wrist rotation, and the varying pressure of the band against the skin. High-end devices use accelerometer data to model and subtract the motion noise; budget trackers do this less effectively.

In published accuracy studies, optical heart rate sensors worn on the wrist typically come within five to ten beats per minute of a medical-grade chest strap at moderate exercise intensities. At high intensity, especially cycling and rowing where the wrist motion pattern is unusual, the error can exceed fifteen beats per minute in either direction. If you are training in specific heart rate zones, a wrist tracker is a useful guide but not a reliable reference. A chest strap electrode or an optical clip sensor at the fingertip is considerably more accurate for that purpose.

Resting heart rate, taken while asleep or sitting still, is where wrist PPG is most reliable. The overnight resting heart rate reading from most major trackers is accurate enough to be a genuinely useful health signal, particularly for tracking trends over time. A single reading means little; a month-long trend that shows resting heart rate climbing unexpectedly is worth attention.

Step Counting and Where It Goes Wrong

Step counting uses the accelerometer. The device measures changes in acceleration and classifies repeated patterns that match the rhythmic motion of walking or running. The classification is handled by firmware algorithms that differ between manufacturers and sometimes between device generations from the same manufacturer.

For walking on flat ground at a normal pace, step counts from most trackers are accurate to within five percent of the actual count. Where the accuracy falls apart is in activities that involve arm motion without steps (chopping vegetables, washing dishes, pushing a trolley), activities with steps but minimal arm swing (pushing a pram or trolley where both hands are occupied), and activities on equipment that fixes the wrist position (a rowing machine, an elliptical, or a bike with handlebars). Putting the tracker on your ankle rather than your wrist improves accuracy for cycling and rowing substantially, though most manufacturers design the interface assuming wrist placement.

Distance calculations derived from steps are less accurate than step counts because they multiply the step count by an estimated stride length. The stride length estimate is either a fixed default or a user-entered height-based calculation. Neither accounts for the fact that stride length changes with speed, gradient, and fatigue. GPS-equipped trackers bypass this problem on outdoor runs by measuring actual distance travelled, but GPS accuracy varies with satellite visibility and is unreliable in urban canyons with tall buildings on both sides.

Sleep Tracking: What the Data Actually Captures

Sleep tracking infers sleep stages from accelerometer data and, in more sophisticated devices, heart rate variability and skin temperature. The accelerometer detects movement; extended periods of stillness are classified as sleep. Heart rate variability is used to distinguish light sleep, deep sleep, and REM sleep on devices that support stage tracking.

The accuracy of sleep stage classification from wrist-worn devices is considerably lower than the accuracy of clinical polysomnography, which measures brain electrical activity, eye movement, muscle tone, and breathing simultaneously. Research comparing consumer trackers to clinical sleep studies consistently finds that trackers overestimate total sleep time and misclassify REM sleep more often than light or deep sleep. The error in total sleep time is typically fifteen to thirty minutes; stage classification is correct roughly seventy to eighty percent of the time for light and deep sleep, and less reliably for REM.

For most people, the clinical accuracy figure matters less than the consistency. A tracker that is wrong by twenty minutes every night in the same direction still tells you whether you slept significantly more or less than usual. Trends are trustworthy even when absolute values are not. Resources like Plur Fitness go into detail on how to use this data practically within a structured fitness plan, particularly around recovery and readiness.

Calorie Estimates: The Least Reliable Number

Calorie burn estimates are the least accurate number that most trackers show prominently. The calculation starts with your basal metabolic rate, derived from your height, weight, age, and sex. It then adds an estimate of active calories based on activity type, duration, heart rate, and movement data. Every step in this chain introduces error, and the errors compound.

Studies comparing tracker calorie estimates to indirect calorimetry (a laboratory method of measuring actual metabolic rate) find errors ranging from fifteen to forty percent, with some devices and some activities producing even larger discrepancies. Cycling is particularly problematic because the wrist motion is limited, making heart rate and movement signals less representative of actual exertion than they are during running.

The direction of error is not consistent. Some trackers overestimate, some underestimate, and the same device can do either depending on activity type. This makes calorie data from trackers unsuitable for precise dietary planning. It is useful as a relative measure, to gauge whether today was a higher or lower output day than yesterday, but should not be treated as an accurate absolute number.

The most useful data from a consumer fitness tracker is resting heart rate trends, sleep duration trends, and activity patterns over time. These three signals, interpreted as trends rather than exact values, are reliable enough to be actionable. The moment-to-moment heart rate during intense exercise and the calorie burn figures deserve considerably more scepticism, particularly if you are making decisions based on them that involve significant effort or significant dietary restriction.

Knowing what the sensors can and cannot do does not make the tracker less useful. It makes the data more useful by directing attention toward the numbers that deserve it.