Sleep trackers can be inaccurate, and there is a surprising lack of testing and research on the subject. This research paper gives an overview of some of the research, and there have been some small studies, like this study in San Diego and this study in Nebraska that compared some commercially available trackers with polysomnography (PSG), which is the gold standard sleep study method (it records brainwaves, heart rate, blood oxygen level, respiration, and movement).

You might reasonably expect sleep trackers with more sensors, capable of recording similar data to PSG, to be more accurate, but manufacturers also develop their own set of proprietary algorithms to interpret the data. With PSG, the data is interpreted by a health care professional.

Some sleep technology companies have funded research. We spoke with Tyler Lee, Asleep cofounder, about some research they plan to present at the World Sleep conference this year. They worked with Stanford and Seoul National University Bundang Hospital on a joint study that compared 11 sleep trackers against PSG. The study examined sleep stage measurement and accuracy in measuring sleep efficiency and latency. The results were interesting.

The study had 75 participants and was conducted independently in Korea. It compared the Pixel Watch, Galaxy Watch 5, Fitbit Sense 2, Apple Watch 8, Oura Ring 3, Withings Sleep Mat, Google Nest Hub 2, and Amazon Halo Rise. It also included three apps: SleepRoutine, SleepScore, and Pillow.

Scores were mixed, but the Pixel Watch, Galaxy Watch 5, and Amazon Halo Rise performed well at measuring sleep stages, while the Oura Ring and Apple Watch were better at measuring sleep efficiency (time in bed spent sleeping) and sleep latency (time to fall asleep). The overall winner was Asleep’s SleepRoutine app, which applies an algorithm to respiration sound. (We are testing the app for a future update, so stay tuned. We’re also testing the Pixel Watch 2.)

The big losers of the study were the other two apps, SleepScore (ultrasound reflection) and Pillow (accelerometer). The Google Nest Hub 2 was also consistently poor, which tallies with our own testing, and the Withings Sleep Mat proved surprisingly bad at determining sleep stages but did quite well with sleep efficiency.