Fitness Trackers: What They Measure and How Well

The core sensor: an accelerometer

Nearly every fitness tracker is built around a small accelerometer, a chip that measures changes in speed along three directions. When you walk, your wrist or hip moves in a repeating pattern, and the accelerometer records it as a wavy signal. Many trackers add a gyroscope, which measures rotation, and some add a barometric sensor, which responds to air pressure and can hint at changes in height.

Those are all examples of the sensor types used across the Internet of Things. None of them measures “a step” or “a workout.” They measure motion, and software has to interpret it.

How steps are estimated

A step counter looks for a repeating peak in the motion signal that fits the rhythm and size of walking or running. A simplified version of the process looks like this:

1. Filter out small vibrations and noise.
2. Look for peaks that rise above a threshold.
3. Check that peaks occur at a plausible pace, often a bit under two per second for walking.
4. Count each qualifying peak as a step.

The weaknesses follow directly. A wrist-worn tracker can miss steps when your arms are still, for example when you push a stroller or carry shopping. It can also count arm movement that is not walking, such as gesturing, brushing teeth or riding over a rough road. A device clipped to a belt or kept in a pocket sees different motion than one on the wrist, so results differ.

Distance, speed and calories

Other numbers are built on top of steps:

• Distance is often step count multiplied by an estimated stride length, based on height and sometimes on how you move. Trackers with satellite positioning can measure distance outdoors from location, which is generally steadier for runs and rides, though tall buildings, tree cover and tunnels can degrade it.
• Speed and pace come from the same sources.
• Calories are estimated from your entered age, weight and height, your movement and, in many devices, your pulse. These are calculations from a model, not measurements of energy burned. Two people doing identical activity can use different amounts of energy, so treat calorie figures as rough.
• Floors climbed typically use air pressure changes, and can be thrown off by weather or by elevators.

Activity types and intensity

Many trackers claim to recognize activities such as walking, running, cycling or swimming. They look at motion patterns, pulse and sometimes location, then use rules or trained models to classify what is happening. This classification works well for common, rhythmic activities and less well for things like weightlifting, climbing, yoga or household chores, where wrist movement does not match the effort. Pulse can help, which is why heart rate sensing on the wrist is often combined with motion. Pulse has its own limits during intense or irregular exercise.

Intensity categories, such as “active minutes” or “zone minutes,” depend on thresholds chosen by the maker. A tracker may count a brisk walk as active while another counts only faster movement, so the totals are not directly comparable between brands.

Why numbers differ between devices

Seeing two different step counts for the same day is normal. Common reasons include:

• Placement. Wrist, waist, ankle, pocket or ring placement changes the motion signal.
• Dominant hand. Wearing on the dominant or non-dominant wrist changes how much arm motion is recorded.
• Different algorithms. Each maker chooses its own thresholds and filters, and updates them over time.
• Sensitivity settings. Some devices count more readily and some less.
• Fit. A loose band moves independently of your arm and adds noise.
• Stride and profile settings. Wrong height or weight entries distort distance and calories.
• Sync timing. Totals may differ simply because one device synced more recently.

The differences are usually a matter of a few percent up to considerably more in awkward situations, but it varies widely and is hard to predict in advance.

How well do they actually do

Independent research on consumer trackers generally finds that step counting is reasonably consistent for ordinary walking, while calorie estimates and some other derived figures are much less reliable. Findings differ by product and test conditions, so be wary of any claim that a device is exactly right. A sensible reading of the evidence is that trackers are good at showing relative change and less good at absolute truth.

Some of the processing happens on the device and some in an app or cloud service. When firmware or an app is updated, the way numbers are calculated can change, which sometimes shifts your history slightly.

Using a tracker sensibly

• Stick to one device when comparing your own days or weeks.
• Wear it the same way each time, with a snug but comfortable fit.
• Enter your personal details accurately and update them when they change.
• Look at weekly patterns rather than a single day’s total.
• Remember that an arbitrary round target is a motivational tool, not a scientific threshold.
• Do not treat estimates as medical data. If you have health concerns or a condition affecting exercise, check with a health professional.

Trackers also raise questions about who sees your activity history. Location traces from outdoor workouts can reveal routines and home addresses, so it is worth reviewing sharing options in the app, a topic covered under wearable privacy. Finally, because trackers have small batteries, features like continuous pulse sampling and satellite positioning will shorten the time between charges.