The program is six weeks of material. What decides the outcome is about five habits, and none of them are interesting.
That is not false modesty about the content. CBT-I is well described in the literature and every decent program teaches roughly the same components, because there is only one body of research to teach from. The variance between people who finish with something and people who do not is almost entirely in how the weeks are run.
Here is what that comes down to.
1. Log every night, especially the bad ones
The plan is not fixed in advance. Your sleep window — the times the program asks you to be in bed — is calculated from what you actually recorded, and recalculated as the weeks go on. A week with four entries produces a plan built on four nights.
The problem with skipping is not only the missing data. It is which nights go missing. People skip the nights they would rather not write down, and those are the bad ones, so a log with gaps reads better than the week was and the window gets set too generously. The bad nights are the ones carrying the information.
It takes under two minutes in the morning and estimates are fine. Nobody needs the exact minute you fell asleep, and trying to know it precisely is its own problem.
2. Hold the rising time, including at weekends
Of everything in the program, this is the one to protect.
A fixed time out of bed is what anchors the whole schedule. It sets when sleep pressure starts building for the following night, and it is the reason a consistent week feels different from an inconsistent one even when the total hours are similar.
Weekends are where it goes. A Saturday lie-in is a small, extremely reasonable decision that shifts the schedule by hours and then takes days to unwind — which is why Monday night is so reliably the worst night of the week for so many people. If you need to move it occasionally, an hour is recoverable. Three is starting again.
3. Do not widen the window ahead of the data
This is the most common way to stall, and it happens precisely because things are going well.
Three good nights arrive, the exhaustion lifts slightly, and going to bed half an hour earlier feels like the obvious reward. What it actually does is dilute the sleep pressure that produced the good nights, and the fourth night is worse.
The window widens on evidence — a week showing consistently efficient sleep — not on a run of nights that felt good. The program works that out for you from the log, which is another reason the log matters. Waiting is unglamorous and it is the whole discipline.
4. Expect week two to be the hard one
Restricting time in bed makes people sleepier during the day before it consolidates sleep at night. This is documented and it is not a side effect of doing it badly — it is the mechanism. Kyle and colleagues measured exactly this: reduced total sleep time, increased daytime sleepiness and measurably impaired vigilance during the restriction phase.
Which means the point at which the program feels least like it is working is the point at which it is doing the thing it does. It is also, unsurprisingly, where most people quit.
