Field note
Five things I changed to sleep better, and how good the evidence is for each
I changed five things: I sleep and wake at the same time, I get sunlight early, I exercise, I keep up with people, and I put my phone outside the bedroom. They worked for me. Writing this page, I found two things I did not expect. The first is that the evidence behind them is wildly uneven, from a study of 88,975 people down to a handful of small ones. The second is that four of the five turn out to be a 1972 clinical protocol I had never heard of, which I arrived at by trial and error fifty years late.
- Regularity has the strongest evidence, and it is not close.
- The phone rule is really stimulus control, formalised in 1972.
- Socialising had research after all. Filed under loneliness.
- My own 28 nights back three of the five.
What this page is and is not
This is a record of what one person changed and what happened, not a protocol. I am a third-year medical student, not a licensed clinician, and n equals one.
What I can offer that a generic list cannot: I know which of these I have actual data on, I know which ones rest on strong research and which rest on almost none, and I am going to tell you which is which rather than presenting five tips as though they were equally supported. They are not equally supported. The gap between the best and worst evidence on this page is enormous.
The five, ranked by evidence
| Change | Strength of evidence | Do I have my own data? |
|---|---|---|
| Same sleep and wake time | Strong. 88,975 people, prospective | No, but it is the one I would defend anyway |
| Phone out of the bedroom | Good, as part of a treatment with the highest AASM rating | Yes: screens before bed cost me 7 minutes of median latency |
| Sunlight early | Strong. A week outdoors moved people 1.4 hours | No |
| Exercise | Moderate and consistently positive | Yes: exercise days were 9 minutes against 15 |
| Seeing people | Moderate. Real findings, filed under loneliness rather than socialising | Indirectly: stress was my single biggest factor |
The uneven column is the honest one. If you only have the energy to change one thing, change the top row.
1. Same time every day
Going to bed and getting up at roughly the same time, weekends included. Of everything on this page, this is the one with real weight behind it, and the finding is more surprising than it sounds.
Windred and colleagues analysed 88,975 UK Biobank participants wearing accelerometers, and built a sleep regularity index scoring how likely you are to be in the same state, asleep or awake, at the same time on consecutive days. The headline: people in the most regular fifth had roughly 30% lower all-cause mortality risk than the least regular fifth, and regularity predicted mortality more strongly than sleep duration did.
That is the part worth sitting with. Almost every piece of sleep advice you have read is about how many hours. This says when may matter more than how long.
The caveats are the same ones I applied to the long-nap mortality studies, and it would be inconsistent to drop them because I like this result better. It is observational. Irregular sleep is caused by shift work, illness, caring responsibilities and poverty, all of which independently affect mortality. Mean age was 62. And it is UK data, where I have leaned American elsewhere on this site. The full picture is here.
What I actually do: a fixed wake time, defended harder than the bedtime. The wake time is the anchor; the bedtime follows it.
2. Phone outside the bedroom
Not face down, not on silent. In a different room, charging where I cannot reach it.
I assumed this was about blue light. It is mostly not. The light effect is real but the exposure in the study everyone cites was four hours of reading on a lit screen, which is not what a two-minute check at 1am looks like. The bigger effect is that a phone within reach turns your bed into a place where you do things, and the single most robust behavioural treatment for insomnia is built on the opposite idea. More on that here.
My own data: across 28 nights in February 2026, nights with a screen in the hour before bed had a median time to fall asleep of 16 minutes against 9 minutes without. That is one person, self-estimated, and those nights overlap with other things. Full caveats here.
What I actually do: phone charges in the next room. The genuinely hard part was the alarm, and the answer was a cheap alarm clock.
3. Sunlight early
Outside, reasonably soon after waking, for a while. Not through a window if I can help it.
The mechanism here is well established: light is the dominant signal that sets your circadian clock, morning light advances it, and the clock is what determines when you get sleepy. That much is not controversial.
The best evidence turned out to be stronger than I first credited. Wright and colleagues at the University of Colorado took eight people camping in the Rocky Mountains for a week with no electric light. Melatonin onset shifted about 1.4 hours earlier, and it happened to every participant, night owls included. That is a large effect, and it came from changing the entire light environment rather than adding a morning dose.
What I still want to be careful about is the leap from that to a prescription. The specific protocols circulating, ten minutes, thirty minutes, within an hour of waking, are far more precise than the evidence I could find behind them. That is the same pattern as the 20-minute nap and the 10-3-2-1-0 rule: a real mechanism wearing a number nobody measured.
The camping study in full, and the evening half that gets ignored.
What I actually do: get outside in the morning, without timing it. I am confident about the direction and not about the dose, so I do not pretend to a dose.
4. Exercise
The evidence that people who exercise regularly fall asleep faster and sleep better is consistent across sources, and the usual caveat is to avoid vigorous exercise close to bedtime.
My own data: days I exercised had a median latency of 9 minutes against 15 minutes on days I did not. That was the third largest gap in my diary. Same warning as always: exercise days were probably also lower-stress days and days I was outside more, and the diary cannot separate them.
What I actually do: something most days, nothing hard in the last three hours before bed.
5. Seeing people
This is the one I believe most strongly from experience and can support least from the literature, and I am putting it last rather than leaving it out.
The case for it is indirect. My single biggest diary gap was stress: high-stress nights had a median latency of 24 minutes against 9 minutes on normal nights, a bigger effect than caffeine, alcohol or screens. Isolation and rumination are stress. A day with actual conversation in it is a day I arrive at bedtime with less unfinished business rehearsing itself.
When I first published this page I wrote that I could not find research connecting social contact to sleep. That was wrong, and it was a shallow search rather than an empty field. It is filed under loneliness, which is why searching for "socialising and sleep" turns up so little.
Cacioppo and colleagues at the University of Chicago found in 2002 that lonely people had poorer sleep efficiency and more time awake after falling asleep, while total sleep duration was unchanged. Loneliness does not shorten the night, it perforates it. And social zeitgeber theory, set out by Ehlers, Frank and Kupfer in 1988, gives the mechanism: social routines act as time cues that entrain circadian rhythms, so a regular social life produces a regular clock.
That connects this row directly to the top row. A week with people in it is a week with a shape, and shape is what the regularity finding is about. The evidence is smaller and more tangled than the accelerometer study, so this belongs in the middle of the table rather than the top, but it is no longer the weakest thing here.
The thing I did not expect
I arrived at these by trial and error over about a year. Writing this page, I discovered four of the five are a clinical protocol from 1972.
Richard Bootzin, an American psychologist, formalised stimulus control therapy for insomnia in 1972. Its instructions include: go to bed only when sleepy, get up at the same time every morning, use the bedroom for sleep only, avoid napping, and get out of bed if you are not asleep after a while. It remains the only nonpharmacological insomnia treatment to hold the highest standard-of-practice rating from the American Academy of Sleep Medicine.
Read that list against mine. Fixed wake time is in it. Bedroom for sleep only is the phone rule. Going to bed only when sleepy is what a regular schedule produces. I had reinvented, badly and slowly, something that has been the standard of care for fifty years.
The lesson I take is not that my changes were unoriginal. It is that the useful advice was already sitting in the clinical literature under a name nobody uses in public, while the internet sold me a great deal of newer, worse-evidenced material with more confident numbers attached. That gap between what is established and what circulates is the whole reason this site exists.
The most famous piece of that protocol, and where its number really comes from, is here.
Questions people actually ask
What is the single most effective change for better sleep?
On the strength of the evidence, a regular sleep and wake time. An analysis of 88,975 UK Biobank participants found that people with the most regular sleep patterns had around 30% lower all-cause mortality risk than the least regular, and that regularity predicted mortality more strongly than sleep duration did.
Does keeping your phone out of the bedroom actually help?
Probably, though less because of blue light than people assume. The stronger reason is behavioural: keeping the bed associated only with sleep is the core of stimulus control therapy, the only nonpharmacological insomnia treatment with the highest AASM standard-of-practice rating. In my own 28-night diary, nights with a screen in the hour before bed had a median latency of 16 minutes against 9 without.
Is sleep regularity more important than sleep duration?
For mortality risk, the UK Biobank analysis found regularity was the stronger predictor. That does not mean duration stops mattering, and the study is observational, so illness and shift work could drive both irregular sleep and worse outcomes.
Does morning sunlight improve sleep?
Yes, and the effect can be large. A University of Colorado study found that one week camping with only natural light advanced melatonin onset by about 1.4 hours in every participant. The specific protocols circulating, such as ten or thirty minutes within an hour of waking, are more precise than the evidence behind them.
What is stimulus control therapy?
A behavioural treatment for insomnia formalised by Richard Bootzin in 1972. Its instructions include going to bed only when sleepy, getting up at the same time each morning, using the bedroom only for sleep, avoiding naps, and leaving the bed if you are not asleep. It holds the highest standard-of-practice rating from the American Academy of Sleep Medicine among nonpharmacological insomnia treatments.
References
- Windred DP, Burns AC, Lane JM, et al. Sleep regularity and mortality: a prospective analysis in the UK Biobank. eLife. 2023. 88,975 participants, sleep regularity index from 7 days of accelerometry. Source for the 30% figure and for regularity outperforming duration.
- Bootzin RR. Stimulus control treatment for insomnia. 1972. Origin of stimulus control therapy and of the instructions quoted here.
- Chang AM, Aeschbach D, Duffy JF, Czeisler CA. Evening use of light-emitting eReaders negatively affects sleep, circadian timing, and next-morning alertness. PNAS. 2015;112(4):1232–1237. Source for the light exposure figures, and for the four-hour exposure caveat.
- Wright KP Jr, McHill AW, Birks BR, et al. Entrainment of the Human Circadian Clock to the Natural Light-Dark Cycle. Current Biology. 2013. Source for the 1.4 hour melatonin advance after a week camping.
- Cacioppo JT, Hawkley LC, Berntson GG, et al. Do Lonely Days Invade the Nights? Potential Social Modulation of Sleep Efficiency. Psychological Science. 2002. Source for loneliness affecting sleep efficiency and fragmentation but not duration.
- Ehlers CL, Frank E, Kupfer DJ. Social zeitgebers and biological rhythms. Archives of General Psychiatry. 1988;45:948–952. Origin of social zeitgeber theory.
- Self-tracking diary, 1 to 28 February 2026. Source for the screen, exercise and stress comparisons. Method and limitations on the nap calculator.
Start with the wake time
If you change one thing, make it a fixed wake time. The calculator works backwards from it and prints every assumption it makes.
Open the sleep calculator