Sleep trackers present detailed stage-by-stage charts each morning. Understanding what the sensors actually record explains which parts of that chart are trustworthy and which are estimates presented with false precision.
The clinical standard measures the brain
Sleep staging in a laboratory is determined from electrical activity recorded at the scalp, alongside eye movement and muscle tone, because the stages are defined by those signals.
Trained scorers assign a stage to each short segment of the night using agreed criteria, and the resulting record is what every consumer device is ultimately compared against.
No wrist or finger device records brain activity, so it cannot measure the quantity that defines what it is reporting.
What the sensors actually capture
A typical wearable measures movement through an accelerometer and blood flow through an optical sensor that detects changes in light absorption at the skin.
From those it derives heart rate, beat-to-beat variability and periods of stillness, then applies an algorithm that maps those patterns onto sleep stages.
The mapping rests on real physiological correlations, since heart rate and variability do differ between stages, but the relationship is loose rather than deterministic.
Duration is measured better than staging
Devices are reasonably good at detecting whether someone is asleep, and total sleep time from a decent wearable is usually close to what a laboratory would record.
They are less good at detecting wakefulness during the night, because lying still while awake resembles sleeping to a movement sensor.
Stage classification is the weakest output, with agreement against laboratory scoring that is well short of what the confident charts imply.
Consistency is worth more than accuracy
An algorithm that systematically over-reports deep sleep will do so every night, which means the direction of change across weeks remains informative.
Comparing a device's absolute numbers against published population figures is where the error matters, and comparing between brands compounds it.
The useful question is whether a wearer's own figures moved after a deliberate change, not whether they match a reference range.
The chart can create the problem
Clinicians report patients presenting with anxiety about their sleep data in the absence of any sleep complaint, having concluded from an application that their sleep is inadequate.
Because anxiety about sleep is one of the more effective ways to disrupt it, the monitoring can generate exactly the difficulty it was bought to detect.
A tracker is a reasonable tool for observing habits and a poor tool for diagnosis, which remains the job of a clinician and, where indicated, a proper sleep study.