Read Your Sleep Journal as a Strip Chart
A sleep diary is only useful if you can see it. Seven lines of times don't tell you that your bedtime moved by two hours across the week; a chart does. This draws your nights the way a chart recorder would, and refuses to claim anything the paper can't show.
- Plotted
- Time in bed per night, mid-sleep drift, recorded wakings, on a 20-hour axis
- local
- Spread maths
- Circular standard deviation — 23:50 and 00:10 are 20 min apart, not 23 h
- stated
- Uploaded
- Nothing — a sleep journal is health data and it stays in this tab
- 0 bytes
- Inferred stages
- None. A diary shows when you were in bed, not what you were doing
- n/a
What this costs you otherwise
| Route | What you get | Cost |
|---|---|---|
| Wearable sleep tracking | Confident stage labels it infers from your wrist | $80–400 |
| Sleep app subscription | Sounds, trends, and a nightly score | $6–12/mo |
| Reading your own diary | Seven lines you can't compare in your head | the pattern |
| This page | $0 |
The 4-step workflow
-
Write down lights-out, not "fell asleep"
You cannot know when you fell asleep from a diary — that's the whole reason the chart is labelled "time in bed". Recording the guess makes every number downstream a guess too.
-
Record wakings as a count and stop there
How many times you woke is roughly right. How long you were awake is not something anyone remembers accurately, so the chart ticks the count and refuses to plot a duration it doesn't have.
-
Look at the two spread numbers before the average
Bedtime spread and wake-time spread, both in minutes. The interesting case is a drifting bedtime with a fixed wake time — that's what produces the weekend of catch-up sleep and the Sunday-night problem.
-
Treat the note tally as a hunch with a table
"Wine on 3 nights, 6.4 h average" is a tally over three nights. The tool prints it and says it is not a correlation. Anything under about thirty nights is a hunch.
The journal format
date | lights out | got up | wakes | notes 2026-09-05 | 23:40 | 06:20 | 1 | wine, phone in bed 2026-09-06 | 01:10 | 08:45 | 0 | out late 2026-09-07 | 23:15 | 05:50 | 3 | hot room Rules the parser follows, and states when it refuses: - a time of 06:20 after a 23:20 bedtime is next morning, not 6pm - 11:30pm and 23:30 both work; "late" and "7ish" do not - 05/09/2026 as a date is ambiguous and is left unplotable - anything under two hours is not treated as a night - unplotable lines are listed, never estimated
What the readings mean
MEDIAN IN BED the middle night, not the average — one 11-hour
weekend won't move it
BEDTIME SPREAD ± minutes around your typical lights-out,
measured on a circle
WAKE-TIME SPREAD ± minutes around your typical alarm
NIGHTS UNDER TARGET your count against your own target, not a
clinical threshold
A regular chart with bad sleep is a different problem from an
irregular chart with the same total hours. The chart shows which
one you have; it does not tell you what to do about it.
Worked example
Sample: what the pre-loaded seven nights chart as Nights plotted 7 Median in bed 7.1 hours Bedtime spread ±43 minutes, measured on a circle Wake-time spread ±53 minutes Recorded wakes 8 across the week Under target 3 of 7 nights, against a 7.5 hour target Uploaded nothing — the journal is plotted in the tab
Numbers above are the output the tool produces from its own pre-loaded example, run in the browser. They are sample data, not averages or measured benchmarks.
What breaks this
Reading the bars as sleep. They are time in bed. Nine hours in bed with two hours awake plots as a long night, and long time in bed with poor sleep is its own problem. Three nights. That's the floor the tool accepts, and it's barely a pattern. Seven shows a weekday shape. Comparing yourself to a wearable. It's inferring stages from movement and heart rate; a diary records what you actually did. They disagree constantly and the diary is the honest one. Rotating shifts scored as bad habits. If you set the rota, the reader describes the work instead of scoring it against a day-shift norm.
Try it now
Four journals are pre-loaded — a week that drifts, a short sleeper with a fixed alarm, a night rota where irregularity is the job, and one written so sloppily that only three nights can be plotted.
Open the strip chart →Go deeper: the trend roll
Free tier charts unlimited journals forever. The $5 one-time pack adds:
- Rolling 90-night chart with week-over-week medians — estimate: ~30 min saved versus stitching seven-day views together to see a trend.
- Note tally across the whole period, so "wine" is counted over 90 nights instead of guessed from one — estimate: ~20 min saved of re-reading your own journal.
- Clinic handout: chart, medians, and your notes on one page — estimate: ~25 min saved of explaining from memory at an appointment.