How is sleep efficiency calculated?

Sleep efficiency is total sleep time divided by time in bed, times 100 (SE = TST / TIB x 100). Total sleep time is the time in bed minus the pre-sleep wait, minus the minutes to fall asleep, minus the time awake mid-night, minus the minutes lying in bed after the final awakening. This explainer animates each subtraction so the number is never a black box.

Why does a sleep diary need across-midnight time math?

Because you get into bed on one calendar day and wake on the next. A lights-out time of 23:15 and a final-wake time of 06:30 span midnight, so the clock reads a smaller number at wake-up. A correct diary adds a full day (1440 minutes) whenever a later event reads an earlier clock time, and rejects impossible sequences rather than showing a negative duration.

Can a sleep diary keep my data private?

Yes, by construction. If the page ships a Content-Security-Policy with connect-src set to none, the browser itself blocks every network request the page could make, so the diary physically cannot upload your sleep data. It is an enforced guarantee rather than a promise. The sleepbook app this explainer describes does exactly that.

Is this explainer medical advice?

No. This is an educational explainer of the arithmetic behind a sleep diary. It does not diagnose, treat, or interpret any score against a clinical threshold. CBT-I for insomnia should be guided by a qualified clinician, and symptoms such as loud snoring or breathing pauses are reasons to see one regardless of any diary.

Sleep Efficiency Explained — How a Sleep Diary Computes SE, TST & SOL

Search “sleep diary template” and you get a page you still do the maths on.

If you have insomnia and you’re doing CBT-I, someone has asked you to keep a diary every morning for weeks. The printables you find are static. You write down six or seven times, then you’re supposed to work out — by hand, half-awake — how long you actually slept and how efficient the night was. Every morning. For a fortnight.

The arithmetic isn’t hard. It’s just fiddly, repetitive, and easy to get wrong — especially the part that crosses midnight. That is exactly the kind of job a page can do for you, correctly, every time.

One night, drawn as a strip.

Picture the whole time you spend in bed as a single bar. The parts where you were asleep fill in jade. The parts where you were awake — waiting to drop off, stirring in the night, lying there after the alarm — stay as dark gaps. That strip is the whole idea.

Add up the jade and you have total sleep time. Take the whole bar and you have time in bed. Everything else is bookkeeping.

How the number is built — one honest subtraction at a time.

Take a real-ish night: into bed 23:00, lights out 23:15, twenty-five minutes to fall asleep, thirty-five minutes awake in the small hours, final wake 06:10, up at 06:40. Here is every step, nothing hidden.

Time in bed 23:00 → 06:40 = 460 min the whole bar
− pre-sleep wait 23:00 → 23:15 = 15 min in bed, not yet trying
− time to fall asleep 25 min sleep-onset latency (SOL)
− awake mid-night 35 min wake after sleep onset (WASO)
− lying awake after waking 06:10 → 06:40 = 30 min early-rise minutes
Total sleep time 460 − 15 − 25 − 35 − 30 = 355 min the jade, added up

SE = TST 355TIB 460 × 100 = 77.2%

That’s it. Sleep efficiency is the share of your time in bed you actually spent asleep. No threshold is applied, no verdict is passed — it’s a measurement, shown with its own formula so you can check it yourself.

The part that trips everyone up: the clock goes backwards at midnight.

Lights out at 23:15, final wake at 06:30. Subtract the clock numbers and you get a negative. A correct diary notices the later event reads an earlier time, adds a whole day, and gets 435 minutes — not a minus sign.

And when the times genuinely don’t make sense — a final wake-up after you got out of bed, or a “night” longer than any real night — the diary refuses to invent a number. It shows a plain error instead of a nonsense one. The screen can never display a negative sleep time.

Your sleep data never leaves the page — and that’s enforced, not promised.

Sleep is health data. The honest way to keep it private isn’t a privacy policy you have to trust — it’s a page that physically cannot send anything. One line of security policy does that:

Content-Security-Policy:
  default-src 'self';
  connect-src 'none';

connect-src 'none' tells the browser to block every outbound request the page could make — no upload, no analytics, no “phone home”, even by accident.

upload to a server blocked
analytics / trackers blocked
external fonts & scripts blocked
save on your device allowed

The diary is stored only in your browser’s local storage. There is no account, no cloud, no share link. Because nothing depends on the network, it also works with no signal at all.

What the diary itself gives you.

The ten-field morning form

The standard sleep-diary items, with inline validation and plain-language errors — a generic implementation of the published fields, not a copied instrument.

Per-night maths, formula shown

SOL, TST and sleep efficiency computed for every night, each printed next to its own formula so nothing is a black box.

The two-week grid

Fourteen nights as stacked strips with 7- and 14-day averages — the exact artifact a clinic asks you to bring.

Trend chart

Sleep efficiency and total sleep time drawn as plain SVG lines across every night. No chart library, no network.

Print, CSV & JSON

A one-page printable handoff for your clinician, a spreadsheet-ready CSV, and a full JSON backup — the only ways data leaves the app.

Nothing it pretends to be

It computes; it never treats. No sleep-restriction schedules, no bedtime prescriptions, no sensor guesses about “true” sleep.

Keep the diary that does the arithmetic.

Enter your mornings, watch the strip and the number appear, print the fortnight when your clinician asks. It stays in your browser — the page can’t send it anywhere.

Open sleepbook →

Free · no account · runs 100% offline