What fitness trackers can and cannot tell you

Wearable devices have become the default way people relate to their own physical activity, and their outputs vary enormously in reliability.

What they measure reasonably well

Validation studies broadly agree.

Step count during walking, which is accurate for most devices in most conditions.

Heart rate at rest and during steady activity, which optical sensors handle reasonably.

Distance and pace during outdoor activity with GPS.

Total sleep duration and timing, approximately.

And relative change over time, which does not require absolute accuracy and is the most useful output.

What they measure poorly

Where the numbers should be treated with scepticism.

Energy expenditure, where validation studies consistently find substantial errors — frequently in the range of twenty to thirty per cent or more, and generally overestimating.

Which matters because people eat back calories the device claims they burned.

Heart rate during high-intensity interval work and resistance training, where wrist-based optical sensors struggle with motion and rapid changes.

Sleep stages, where no consumer device measures brain activity and agreement with laboratory measurement is poor.

Step counting during activities other than walking.

Blood pressure and blood glucose from optical sensors, which are not validated clinical measurements in most consumer devices.

And proprietary composite scores — readiness, recovery, body battery, fitness age — which are unvalidated algorithmic constructs rather than measurements.

Where they genuinely help

The behavioural evidence.

Trials of activity monitors generally find modest increases in physical activity, sustained better when combined with goal setting, feedback and support.

The mechanism is awareness and accountability rather than physiological insight.

They are useful for identifying patterns — which days are inactive, how much sitting occurs, whether sleep is regular.

And they provide a record, which counters the sense of doing nothing.

Where they cause problems

Documented in the literature.

Anxiety and preoccupation with metrics, including orthosomnia — sleep problems generated or worsened by tracker scores.

Guilt on days when targets are not met, which for some people reduces activity rather than increasing it.

Over-reliance on calorie figures for weight management, which are unreliable.

Ignoring bodily signals in favour of the device, particularly in training and recovery.

And false reassurance from unvalidated health features.

The health features worth knowing about

A more nuanced picture.

Irregular rhythm notifications, which have been validated in large studies for detecting atrial fibrillation and which have led to genuine diagnoses.

They also generate false positives and considerable anxiety, and a notification requires medical confirmation with a proper recording rather than being either dismissed or acted on alone.

Single-lead electrocardiogram features on some devices, which are regulated as medical devices in some jurisdictions and which have defined limitations.

Fall detection, which has genuine value for older users.

Blood oxygen measurement, which is not a clinical-grade measurement in most consumer devices.

And screening features in general, which apply to a population where the condition is uncommon, meaning most positives are false.

Using one well

A reasonable approach.

Use it for consistency and trends rather than absolute numbers.

Ignore the calorie figure for dietary purposes.

Ignore sleep stage breakdowns.

Treat composite scores as entertainment.

Set targets you will meet on a bad day rather than aspirational ones.

Take it off periodically, particularly if it has become a source of anxiety.

And use the data to answer specific questions — does alcohol affect my sleep, am I more active on days I cycle to work — rather than monitoring everything continuously.

Data and privacy

Worth considering.

Health and location data collected by consumer devices is not always covered by medical data protections, and terms vary considerably.

Data has been shared with third parties, used in insurance products and, in some documented cases, exposed publicly through aggregated features.

Reading what is collected and shared, and adjusting settings, takes a few minutes.

And insurance-linked activity schemes involve a genuine trade-off between discounts and data that is worth making deliberately.

Whether to buy one

An honest assessment.

If you will use it to build a habit and you are the kind of person who responds to feedback, it is a reasonable purchase.

If you already exercise consistently, it adds little.

If you are prone to anxiety about numbers, it may make things worse.

A phone counts steps for free, which covers the most useful output.

And no device is required to walk more, which remains the intervention with the best evidence and the lowest cost.

General information only, not medical advice. Consumer devices do not diagnose conditions — consult a qualified clinician about any health concern or device notification.