Sleep efficiency

The arithmetic is the whole definition. Take the minutes you actually slept, divide by the minutes you were in bed, multiply by 100. Everything interesting about sleep efficiency is a question about what went into the numerator and what fell out of it.

A worked night: you are in bed for eight hours, so 480 minutes of time in bed. It takes 20 minutes to fall asleep, you are awake for 25 minutes somewhere in the middle, and you lie there for 15 minutes at the end before getting up. That is 60 minutes lost and 420 minutes of total sleep time, so 420 divided by 480 is roughly 88%. Around 85% or higher is the conventional benchmark for adults. It is a benchmark, not a health grade, and one night underneath it means nothing.

What drags sleep efficiency down

Only three things can. A long sleep latency at the start, wake after sleep onset in the middle, and awake time at the end before you get up. That is more useful than it sounds, because it turns a vague sense of a bad night into a specific question about which of the three is eating it, and each one has a different answer.

The trap is the denominator. Efficiency falls whenever time in bed grows, so the instinct to go to bed an hour earlier after a rough week usually pushes the number down rather than up: you gain a little sleep and a lot of lying awake. Clinicians treating chronic insomnia sometimes do the reverse deliberately, compressing time in bed to concentrate sleep and then widening it again, but that is a supervised technique and not something to improvise from a definition page.

Tock computes efficiency for each tracked night and shows it next to the parts it is made of, so a low percentage points somewhere rather than just looking bad. For what a phone can and cannot see while it does that, read sleep tracking without a wearable.