The night my trackers disagreed

This morning, for the first time, I pushed back on what my trackers told me.

I wear two trackers. Yes, two, and I sleep in both of them. Last night one of them reported that I had been awake for two hours in the middle of the night. The other failed to pair and had nothing to say. On most mornings the two devices and my own sense of things form a small committee, and the devices get the deciding vote. This morning the committee was short a member.

So I did the thing I have spent a year building an app to help people do. I stopped and asked my body a direct question: am I actually tired, or do I feel like someone who was told she slept five hours? I sat with it for about thirty seconds. The answer was that I was rested. I was not wired and I was not foggy. I was rested.

I went to morning yoga. I am fairly sure that if the second tracker had paired and agreed with the first, I would have skipped it, because two devices saying you were up for two hours is very persuasive at six in the morning. Instead I am more awake and more settled than I have been most mornings this month, and I owe that to a question that took half a minute.

Afterward I went looking for what the research says about the number I almost believed, and it changed how I read these reports.

The part of a sleep report I had been treating as fact is the part the devices are worst at, and this is true of the current generation, not just the early ones. In 2025, a team at Hasselt University and Antwerp University Hospital tested six wrist trackers on sale today, among them the Apple Watch Series 8, the Whoop 4.0, two Fitbits, a Garmin, and a Withings, against polysomnography, the laboratory standard. Every device recognized sleep more than 90 percent of the time. At correctly recognizing wake, the best of them managed 52 percent and the worst 29 percent, and every one of them underestimated the time a person spent awake during the night, by 12 to 48 minutes. (Schyvens et al., 2025, SLEEP Advances) A 2024 study at Brigham and Women’s Hospital, funded by Oura, found the same shape with the Oura Ring Gen3, the Fitbit Sense 2, and the Apple Watch Series 8: at least 95 percent on sleep, and between 52 and 69 percent on wake. (Robbins et al., 2024, Sensors) The field does not even agree on which direction the error runs. When a 2025 meta-analysis pooled 24 studies and 798 people, wrist trackers as a group overestimated night-time wake by about 13 minutes. (Lee et al., 2025, Journal of Clinical Sleep Medicine) Some devices miss wake, some invent it, and the number I was about to plan my morning around is the least settled number on the screen.

Sleep clinicians have a name for what I was doing. In 2017, Kelly Baron and colleagues at Northwestern published a paper in the Journal of Clinical Sleep Medicine about patients who arrived seeking treatment for sleep problems they had diagnosed from their tracker data. They called it orthosomnia. One patient, after a night in the sleep lab showed she was getting deep sleep, asked, “Then why does my Fitbit say I am sleeping poorly?” The authors wrote that consumer trackers “are unable to accurately discriminate stages of sleep and have poor accuracy in detecting wake after sleep onset.” (Baron et al., 2017, Journal of Clinical Sleep Medicine)

And being told you slept badly does something on its own. In 2014, Christina Draganich and Kristi Erdal ran an experiment with 164 students. Each student was told, by what looked like a measurement, either that they had spent 28.7 percent of the night in REM sleep, which is above average, or 16.2 percent, which is well below. The numbers were invented. The students who were told they had slept poorly did worse on a test of attention and information processing, and their own sense of how they had slept did not predict their score. The researchers called it placebo sleep. (Draganich and Erdal, 2014, Journal of Experimental Psychology: Learning, Memory, and Cognition)

Put those together and the morning makes sense. A device with a poor record on wake told me I had been awake for two hours, and if I had believed it, the belief alone could have made me perform like someone who had. The only instrument in the room with a good record on how I actually felt was me, and it had not been asked.

Here is what I believe now. The numbers are a second opinion, and my body gets the first one. I still wear the trackers, and I still look at them. Their measurements are useful for what they are good at, which is patterns over weeks. They are not good at telling me how I am this morning, and I had been letting them answer that question for years.

If you wear a tracker, I would like to know what happens on the mornings when it disagrees with you. Do you check the number before you check yourself, and which one wins?