How I Rewired My Evening Routine After Realizing It Was Sabotaging My Sleep
The Night I Looked at My Sleep Data and Felt Embarrassed
I had been tracking my sleep with an Oura ring for eight months. My average sleep score: 62. I blamed it on stress, on a demanding job, on the general ambient noise of life.
Then one night I actually looked at what time I was going to bed versus what time my phone usage patterns suggested I was settling down, and I noticed the problem immediately. I was technically “in bed” by 10:30pm, but I was on my phone until 11:45pm, eating something small around 10, having my last coffee at 5pm but often a glass of wine at 8pm, and keeping the overhead lights at full brightness until the moment I turned them off and expected to sleep.
I was sending my nervous system contradictory signals for hour. Then wondering why sleep onset took 30+ minutes and why my deep sleep was shallow.
This is the story of what I changed, why it worked, and what the science says about each piece.
The Before Picture
| Behavior | Typical Time | Sleep Impact |
|---|---|---|
| Last coffee | 5:00pm | Caffeine half-life: still 25-30% active at 10pm |
| Dinner | 8:30-9:00pm | Late meal raises core temp and insulin near sleep |
| Alcohol | One glass around 8pm | Disrupts REM in the second half of the night |
| Overhead lights | Full brightness until 10:30pm | Suppresses melatonin production |
| Phone use | 10:30pm - 11:45pm | Blue light + cortisol-triggering content |
| “Bed time” | 10:30pm | Theoretical; actual wind-down: midnight or later |
| Average sleep score | 62 / 100 | Oura ring, tracked over 8 months |
Almost nothing in that list was helping sleep. Most of it was actively working against it.
What I Changed (And Why Each Change Matters)
Change 1: Moved the last coffee to 1pm
Caffeine has a half-life of approximately 5-6 hours. A coffee at 5pm means 25-30% of that caffeine is still circulating at 10pm, and it does not stop being adenosine-blocking (the mechanism of caffeine’s alertness effect) just because you have stopped noticing it.
Research published in the Journal of Clinical Sleep Medicine found that caffeine consumed even 6 hours before bedtime significantly reduced total sleep time. Moving my last coffee to 1pm was uncomfortable for approximately one week and then no longer noticeable.
Change 2: Dinner finished by 7pm
Late eating raises blood glucose and core body temperature simultaneously - both of which are obstacles to sleep onset and deep sleep quality. Core body temperature must drop roughly 2-3 degrees Fahrenheit to initiate deep sleep, and digestion keeps it elevated.
In my CGM data (I wore a continuous glucose monitor for 30 days around this period), finishing dinner by 7pm produced a glucose profile that had fully stabilized before bed. My Oura deep sleep average improved by 14 minutes within the first two weeks.
Reference: Sleep Foundation on nutrition and sleep timing
Change 3: No alcohol on weeknights
This was the change I resisted longest and the one that produced the clearest, most immediate improvement.
Alcohol disrupts REM sleep in the second and third sleep cycles, which typically occur in the second half of the night. You may fall asleep faster (alcohol is sedating), but research from the University of Melbourne showed that even moderate alcohol - less than two drinks - reduced REM sleep by 24% and increased nighttime disturbances significantly. My Oura data showed exactly this pattern every time I drank, even a single glass.
The 1-2am wakeups I had attributed to stress stopped almost immediately once weeknight alcohol was removed.
Change 4: Dimmed lights and amber bulbs after 7:30pm
Light signals the SCN (suprachiasmatic nucleus - your circadian clock) about the time of day. Bright overhead lighting after sunset tells your brain it is still day and suppresses melatonin production. Research from Harvard Medical School found that exposure to room light in the hours before bed suppresses melatonin onset and shortens melatonin duration - effectively delaying sleep readiness.
I installed smart bulbs set to shift automatically to low-intensity amber from 7:30pm onward. Blackout curtains went up. The visual environment now communicates evening to my nervous system rather than fighting against it.
Change 5: No phone after 9pm
This was less about blue light (which is real but overstated as a mechanism) and more about content. Checking news, email, or social media before bed activates the analytical and reactive parts of the brain, increasing cognitive and physiological arousal and making the cognitive shutdown required for sleep onset significantly slower.
I put the phone in another room at 9pm. I bought a cheap analog alarm clock. The first week involved conscious discomfort. By week three it was simply habit.
Change 6: A structured wind-down sequence (9-10pm)
The replacement for phone time became a repeatable sequence:
- 20 minutes of reading (physical book only)
- A 10-minute warm shower or bath (the subsequent body temperature drop accelerates sleep onset according to research in Sleep Medicine Reviews)
- 5-10 minutes of light stretching or progressive muscle relaxation
- In bed by 10pm, lights out by 10:10pm
The sequence itself matters less than the consistency. A repeatable pre-sleep sequence functions as a conditioned signal - your nervous system learns to associate the sequence with sleep, making the transition faster over time.
The After Picture (12 Weeks Later)
| Metric | Before | 12 Weeks After | Change |
|---|---|---|---|
| Sleep score (Oura avg) | 62 | 79 | +27% |
| Sleep onset (min) | 31 | 9 | Faster by 71% |
| Deep sleep (min/night) | 44 | 67 | +52% |
| REM sleep (min/night) | 88 | 108 | +23% |
| Night wakeups | 2.9 | 0.7 | -76% |
| Morning readiness score | 61 | 78 | +28% |
No supplements added. No sleep aids. No tracking devices beyond the one I already had. Six behavior changes, all of which cost nothing.
The One Change to Start With
If this list feels overwhelming, pick the one with the highest likely payoff for your specific situation. For most people it is one of these three:
- Last alcohol 3+ hours before sleep (most impactful for people who drink regularly)
- Phone out of bedroom by 9pm (most impactful for people who scroll in bed)
- Lights dimmed after 7:30pm (most impactful for people with long sleep onset times)
Start with one. Track your sleep score or simply how you feel in the morning for two weeks. If it moves, that one was the problem. If it does not, try the next.
The evening is what the morning is built on. Most people optimize the morning without touching the evening, and wonder why the mornings keep underperforming.
