The column in my spreadsheet that I looked at most was not the bedtime. It was the one next to it, the running total of hours I had missed that week. A bedtime I could argue with. A total I could not.
This is that column, with the payback worked out. Give it the hours you slept on each of the last seven nights, or one average if you honestly cannot remember, and it adds up how short you are. Then it shows two ways to pay it back: a bit earlier every night, or 90 extra minutes on two nights a week.
How the math works
Start with what you need. The calculator sets it to 8 hours a night, and the slider runs from 7 to 9, because that is the range the National Sleep Foundation panel settled on for adults (Hirshkowitz et al., Sleep Health, 2015). If you know you are a 7-hour person, say so. If you are honestly not sure, leave it at 8.
Then, for each night, it takes the need and subtracts what you slept. A short night adds the difference. A night at or over the need adds zero. Not a negative number, zero.
That last rule is the one people argue with me about, so here is why. Recovery sleep is real. Sleep longer after a bad week and you do get some of it back. But it is not one for one, and I could not find a study that says a 10-hour Saturday erases a 6-hour Tuesday. What I did find was Kitamura and colleagues (Scientific Reports, 2016), who took people carrying roughly an hour a night of debt and let them sleep as long as they liked. It took several nights of longer sleep before their sleep settled, not one. So the sum counts the short nights and lets the long ones be a nice weekend.
The nightly plan divides the debt by the extra sleep you are willing to add each night, rounded up, because half a night is not a thing. The extra is set to 60 minutes and the slider goes from 30 to 90. It also moves your lights-out earlier by that amount, so if you usually turn the light off at 11:00 p.m. and pick 60 minutes, the plan says 10:00 p.m. and tells you for how many nights.
The weekend plan is for people who cannot move a weeknight. It gives you 90 extra minutes on two nights a week, works out how many weekend nights that takes, and then how many weekends. The first week on the switchboard I was 9 hours down by Sunday and slept until noon to fix it. It did not fix it, and I had to work off the rest a night at a time.
What the sum can and cannot tell you
The sum is good at one thing: the last seven nights. That is why the calculator takes seven and not thirty. Van Dongen and colleagues (Sleep, 2003) kept a group at 6 hours a night for two weeks. Their reaction times got worse every day of it, and kept getting worse, while the people themselves stopped reporting that they felt sleepier after the first few days. What you feel is not the whole debt. It also means an old debt, the one from last month, is not something adding up seven nights can see, and I would not pretend it can.
A large number does not clear in one long night. Everything I have read on recovery says it takes several nights of longer sleep, and my own spreadsheet agrees. If the calculator gives you 20 hours, that is a rough month, and the 14 nights at 90 minutes it suggests are the plan. Not a sentence, a plan.
To find the earlier lights-out that also ends a sleep cycle at your alarm, put the new bedtime into the sleep calculator. I wrote about why the cycle matters in what time should I go to bed.
One more thing, and I say it on every page. If you are getting enough time in bed and still feel wrecked, if you snore hard enough to wake yourself, or if you lie awake for an hour most nights, that is not a debt problem and this sum cannot see it. Those are questions for a doctor. I am not a doctor, and I would rather say so plainly than have you rearrange a month of bedtimes around a number that is not the problem.
Tonight, then. Type in the last seven nights as honestly as you can, look at the number, and move the light off by the amount the nightly plan says. Just tonight. Tomorrow it will say one night fewer.