Why Sleep Trackers Disagree

What is measured and what is estimated

• Measured directly by the device: movement, and on many devices a pulse signal from light on the skin. Some bedside devices sense movement and breathing without touching you.
• Estimated by an algorithm: when you fell asleep, when you woke, total sleep time, and "light", "deep" and "REM" stages.
• Calculated by the maker: a sleep score or readiness number. These have no shared definition between brands and no universal clinical meaning.

The reference method in research is polysomnography (PSG), an overnight lab recording that includes brain waves. A framework paper on testing sleep trackers notes that, because the industry moves fast and testing has not been standardized, the accuracy and reliability of these devices in measuring sleep remains largely unknown. [1] The American Academy of Sleep Medicine has said that consumer sleep technology cannot be used for diagnosis or treatment of sleep disorders. [2]

What studies found

1. Sleep, 2021

   Detecting sleep was good, detecting wake was weak to moderate, and stage results were mixed. [3][4]

   Who and what: 34 healthy young adults, three nights in a lab; four wearable and three bedside devices

2. JMIR, 2023

   Stage agreement varied widely between devices. [5]

   Who and what: 75 people in two Korean clinics; 11 devices

3. Sleep Advances, 2026

   Devices may be less accurate in older adults. [6]

   Who and what: 13 young and 19 older adults, one night; two wearables and two bedside devices

4. Sleep, 2026

   Between-device agreement was weak for most measures. [7]

   Who and what: 74 older adults at home, four modalities

Why the numbers differ

1. Different sensors. In the Korean study, wearables and bedside devices showed high proportional bias in different measures (sleep efficiency for wearables, time to fall asleep for bedside devices). [8]
2. Wake is hard to see. In the 2021 lab study, the devices were very good at calling sleep "sleep" but much worse at recognizing wake, and they tended to do worse on poorer or disrupted nights. [3][9] A quiet, still person lying awake can look asleep to a movement-based device.
3. Stages are the least reliable part. In the 2026 study of young and older adults, devices performed poorly at identifying individual sleep stages, particularly deep sleep. [10] A home study of older adults also found poor agreement on stage durations. [11]
4. People differ. In the Korean study, accuracy varied with factors such as BMI, sleep efficiency and the apnea-hypopnea index. [12] Results from healthy young volunteers may not hold for older adults. [6]
5. Different questions. In the home study of older adults, duration was the only aspect of sleep that moved consistently between devices. [13]

How to use this

• Compare a device against itself over weeks, not against another device on one night. See Sleep Trends vs Individual Nights.
• Treat total time asleep and bed and wake times as the most useful numbers, and stage and score numbers as rough. See Sleep-Stage Estimates: How Accurate Are They?.
• Add how you felt. A short diary often tells you something a device cannot. See Sleep Diaries vs Devices.
• For the gold-standard comparison, see Wearables vs Sleep Studies.
• Accuracy is not the same as benefit. A device that tracked sleep perfectly would still not prove that tracking improves your sleep or health.

When to talk with a clinician