I was sitting on my kitchen floor at 3:47 AM, crying into a bowl of cereal I didn't even want, when I realized I couldn't do this anymore. I'd slept maybe ninety minutes. My sleep tracker said I'd been awake since 1:15. Before that, I'd lain rigid in bed from 10 PM to 12:30 AM, watching the ceiling fan, feeling my heart pound every time I checked the clock.
That was April 3rd. The night before I started CBT-I.
I'm not a doctor. I'm just someone who spent two years averaging about four hours of broken sleep a night, tried every supplement and sleep hygiene tip on the internet, and finally got desperate enough to try the one thing I'd been avoiding because it sounded too hard: Cognitive Behavioral Therapy for Insomnia.
This is what those six weeks actually looked like. The numbers. The setbacks. The morning I woke up and realized something had changed.
Week 1: The Sleep Window (Sleep Efficiency: 62%)
The first thing I did was calculate my sleep window using my two-week baseline average. I was in bed about eight hours but sleeping maybe three and a half. My sleep efficiency was hovering around 44%. The protocol was clear: match your time in bed to your actual sleep time, then add thirty minutes.
So my sleep window became 12:30 AM to 4:30 AM. Four hours.
I stared at that number for twenty minutes before I accepted it. Four hours felt like admitting defeat. But I was already only getting that much anyway—I was just spending the other four hours marinating in anxiety.
The first three nights were brutal. Not because I was more tired—I'd been exhausted for months—but because staying up until 12:30 when every cell in my body was screaming to go lie down and try required a kind of discipline I didn't know I had. I deep-cleaned my bathroom. I folded laundry in slow motion. I sat on the couch and watched the clock.
But here's what happened: when 12:30 came, I was asleep within fifteen minutes. And I stayed asleep until 3:45. Then I was up. No falling back. At 4:30, I got out of bed even though I wanted to bargain for "just twenty more minutes."
Three and a half hours. Sleep efficiency: 62%. It was working, sort of, and I hated it.
Week 2: The First Crack (Sleep Efficiency: 71%)
By day eight, something shifted. I fell asleep at 12:28 and woke up at 4:15. Three hours forty-seven minutes. But it was solid. I didn't wake up to pee. I didn't wake up to check my phone. I just… slept.
I expanded my window by fifteen minutes based on the protocol: 12:15 AM to 4:30 AM.
On day ten, I slept from 12:13 to 4:25. Four hours twelve minutes. I cried in the shower that morning—not sad crying, just release. Dr. Charles Morin at Université Laval, one of the researchers who helped develop CBT-I protocols, talks about "re-establishing sleep pressure." That's what was happening. My body was finally learning: bed equals sleep, not bed equals worry.
I was still a zombie during the day. I won't lie about that. But it was a different kind of tired. Not the wired, panicky exhaustion. Just… sleepy.
Week 3: The Relapse Night (Sleep Efficiency: 68%)
Day sixteen. I went to bed at 12:15 and lay there until 2:00 AM, wide awake, chest tight, thoughts racing. The old terror came back: it stopped working. I broke it. I'm going to be awake forever.
I wanted to get up, but I didn't. I broke the stimulus control rule—one of the core CBT-I principles is that if you're awake more than fifteen to twenty minutes, you leave the bedroom. I just couldn't move. I was too scared to "reset" the night.
I finally fell asleep around 2:20. Woke up at 4:30 to my alarm. Two hours ten minutes.
I wanted to quit. I wanted to go back to melatonin and magnesium and sleeping in until whenever. But the protocol said: one bad night doesn't erase progress. Don't change your window based on a single night. So I didn't.
The next night, I slept 12:18 to 4:28. Four hours ten minutes. My body was still learning.
Week 4: The Turning Point (Sleep Efficiency: 78%)
I expanded my window to 12:00 AM to 5:00 AM. Five hours in bed.
On day twenty-four, I slept straight through. Midnight to 5:00 AM. Five full hours. I woke up before my alarm and just lay there, feeling my body, noticing that I didn't feel dread. Just awake. Calm.
That week, four out of seven nights hit 80%+ sleep efficiency. The math became a game instead of a judgment. I started trusting the process—the boring, repetitive, undramatic process of going to bed at the same time and getting up at the same time, no matter what.
I also started catching the thoughts. The ones that would spiral at 2 AM: If I don't sleep tonight, tomorrow is ruined. I'd write them down in the morning, look at them in daylight, see how catastrophic and vague they were. It didn't make them disappear, but it made them less sharp.
Week 5: Consolidation (Sleep Efficiency: 83%)
Window: 11:45 PM to 5:15 AM. Five and a half hours.
This was the week I stopped white-knuckling. I stopped checking my sleep tracker first thing in the morning. I just got up, showered, made coffee. If I'd slept well, great. If I hadn't, okay—tonight's another chance.
I had two nights over 85% efficiency. One night at 91%. I was averaging about four and a half hours of actual sleep, but it was deep. Restorative. I started noticing things: I could focus in meetings. I didn't need three coffees before 10 AM. My shoulders weren't constantly up around my ears.
Week 6: Sleeping Through (Sleep Efficiency: 87%)
Window: 11:30 PM to 5:30 AM. Six hours.
On day thirty-nine, I went to bed at 11:28 PM and woke up at 5:29 AM. Six hours, one minute. Straight through. No waking. No panic. No checking the clock.
I didn't even realize it until I looked at my log that morning. I'd just… slept.
By the end of week six, I was averaging five hours forty minutes of sleep per night, with an efficiency hovering between 85% and 90%. The protocol suggested I could start expanding toward six and a half or seven hours in bed, fifteen minutes at a time, as long as efficiency stayed above 85%.
I still have nights where I wake up at 3 and need to use the stimulus control rule—get up, sit in the dim hallway, go back when I'm drowsy. But those nights don't scare me anymore. They're just data. Just part of being human.
If you're where I was in April—exhausted, desperate, willing to try something structured even if it sounds hard—I used a guide called Rest →. It's the six-week CBT-I protocol I followed, with the sleep window calculator, the logs, the cognitive work. It's not magic. It's just a system. But it worked when nothing else did.
I'm writing this at 10:47 PM. I'll go to bed in about forty minutes. I'll probably sleep. And if I don't, I'll get up at 5:30 anyway, and I'll be okay.
That's the part I didn't expect: not just sleeping better, but losing the fear of not sleeping.
— Simon