It is 2am. You are not asleep. You are holding a phone that is, at this exact moment, recording that you are not asleep.
The schedule slipped weeks ago. Bedtime drifted later, then untethered. Now you sleep at 4 and wake at noon and feel hungover from rest you technically got. You downloaded an app to fix it. It rates your nights. It also quietly builds a profile of one of the most intimate signals a body produces: when it shuts down and when it comes back. Nobody asked for that part.
The privacy problem with sleep tools specifically
Sleep data is not neutral. Bedtime, wake time, the hours you are reachable and the hours you go dark, these reveal shift work, depression, a newborn, a breakup. A record of when you are unconscious is close to a record of when you are vulnerable.
Most sleep and wellness apps keep that record. Some have been caught monetizing the adjacent data. GoodRx paid a USD 1.5M FTC penalty in February 2023 for sharing health data with Google and Facebook for advertising (source: FTC, February 2023). The category pattern is the same: the product is cheap, and the data is the asset.
Shift does not keep this record. The server is stateless. No signup, no account, no cookies, no logs. Nothing about your conversation is stored after it ends, so there is no nightly time series of your sleep on a disk anywhere. You can confirm it from view-source. There is no breach risk to your sleep data here because there is no data. You fix the behavior, and the tool remembers nothing.
What actually moves a broken schedule
A wrecked schedule is rarely an insomnia problem. It is a timing problem. Your internal clock has drifted off the sun, and three levers pull it back. None require a wearable.
Anchor the wake time first, not the bedtime. You cannot force sleep at 11pm by deciding to. Fix a wake time and hold it, including weekends, and let the bedtime follow it back into place. The anchor is the morning, not the night.
Get bright light early and dim light late. Light is the strongest input to the clock. Ten to twenty minutes of outdoor light within an hour of waking tells your body the day started. Cutting bright light in the last hour before bed stops telling it the day is still going. This is the highest-leverage change for most drifted schedules.
Build a wind-down that is short and the same every night. Thirty minutes, repeatable, boring on purpose. A consistent signal that the day is closing, so your body stops waiting for one more thing.
These are slow. A drifted clock moves back about an hour a day at best. The fix is days, not one heroic night.
How the Shift method handles this
You already half-know the list above. Knowing it is not the problem.
The method is built on Motivational Interviewing and the Transtheoretical (stages of change) model, the same ground laid out in the method. It starts by finding where you actually are. Someone running a 4am bedtime on caffeine and fear is not failing to understand sleep hygiene. The schedule is usually doing a job: late nights are the only unclaimed hours, or sleep is where the anxious thoughts arrive, or being awake feels like control. Shift asks what the late night is protecting before it asks you to give it up.
You can see that pattern in a synthetic session with a founder convinced he could not sleep more without losing his company. The work was not teaching him sleep facts. It was examining the belief under the schedule.
An honest note on limits
This page is about a drifted schedule and the behavior that resets it. It is not medical advice.
If you cannot sleep despite a stable schedule and good light, if you snore heavily and wake unrefreshed, or if the sleeplessness comes with dread that does not lift, that is a medical question, not a habit one. Talk to a doctor. Sleep apnea, thyroid problems, and depression all wear the costume of a bad schedule.
And if the nights are dark in the other sense, if you are awake at 2am because staying alive feels like the hard part, that is not for any app. In the US, call or text 988, the Suicide and Crisis Lifeline. A real person will answer.