How to Track Your Mood (And Actually Keep It Up)

6 September 2026 · 6 min read

Most people who start a mood journal stop within two weeks. Usually not from lack of interest, but from having designed something nobody could sustain. Here is what to record, when, and what to expect.

Why bother recording it at all

The honest case for mood tracking is not that writing things down is inherently therapeutic. It is that human memory is a bad instrument for measuring your own emotional life, and it fails in specific, predictable ways.

We remember peaks and endings far better than averages. Ask someone how their month was and the answer will be disproportionately shaped by the worst afternoon in it and by whatever happened in the last few days. Ask how often they felt anxious and you will get a number derived from how vividly anxious episodes come to mind, not how frequently they occurred. A week with one terrible Tuesday and six unremarkable days gets filed as "a bad week".

A record fixes this cheaply. Not by being profound, but by being the only version of events that isn't reconstructed after the fact.

There is a second, smaller effect. Research on affect labelling, putting feelings into words, suggests the act of naming an emotion changes how it lands. In a well-known 2007 study, Matthew Lieberman and colleagues found that labelling an emotional image was associated with reduced activity in the amygdala. Naming a feeling is not the same as resolving it, but "I am anxious about Thursday" is a more tractable object than a formless bad mood.

The two ways this goes wrong

Nearly every abandoned mood journal failed in one of two directions.

Too much detail

A form with nine sliders, a free-text box, sleep hours, caffeine, exercise and a weather field is a genuinely interesting dataset and takes four minutes to fill in. Four minutes is far beyond what anyone sustains daily for months. The entry becomes a small chore, the chore gets postponed, and postponed entries get invented later, which is precisely the reconstructed-memory problem you were trying to escape.

Too little detail

The opposite failure is a single number with no context. Sixty days of "6, 7, 4, 8, 5" tells you your average and nothing else. You cannot act on it, because there is nothing attached to explain why the 4s were 4s. It sustains easily and teaches you nothing, so eventually you stop seeing the point.

What to actually record

A rating you can use consistently

Five points is usually right. Ten-point scales feel more precise but people cannot reliably distinguish a 6 from a 7 in their own mood, so the extra resolution is noise dressed as data. What matters is that "3" means roughly the same thing to you in November as it did in September.

One or two tags

This is the part that makes the numbers mean something. Tags are what let you eventually notice that your low days cluster around short sleep, or long stretches without seeing anyone, or particular kinds of work. Keep the list short: six to ten tags you'd actually use. A taxonomy of forty is a taxonomy you'll ignore.

A sentence, when there's one worth writing

Optional is the important word. On most days there is nothing to add and the rating plus tags is the whole entry. On the days when something happened, one line of context is worth more than a fortnight of ratings, because it is what you'll actually recognise when you look back in three months.

Naming the feeling more precisely helps here. "Bad" covers frustrated, lonely, exhausted, and apprehensive, which have almost nothing in common and call for different responses. Research on emotional granularity, associated particularly with the work of Lisa Feldman Barrett, suggests that people who make finer distinctions between emotional states tend to regulate them more effectively. You don't need a vocabulary list, just the habit of asking which of several bad things this actually is.

When to log

Pick one time and keep it. Consistency matters more than which time you pick, because entries recorded at wildly different points in the day aren't comparable to each other.

Evening has the advantage of covering the whole day and the disadvantage of being filtered through how the day ended, since a good last hour lifts the rating for a mediocre day. Morning is cleaner but reports on yesterday. Either works. Drifting between them doesn't.

Attach it to something that already happens at that time rather than relying on a reminder you'll learn to dismiss. After brushing your teeth, after putting your phone on to charge, after the last email of the day. This is the same mechanism as habit stacking, applied to the tracking itself.

How long before it tells you anything

Two to four weeks before anything meaningful appears, and there's a mathematical reason rather than a motivational one.

Most of the patterns worth finding are weekly. If Sunday evenings are reliably harder, that pattern needs three or four Sundays before it separates from coincidence. Ten days of data cannot distinguish "Mondays are hard" from "that particular Monday was hard". So the first fortnight will feel like filing paperwork into a void. That is the correct experience of the first fortnight.

What tends to surface once there's enough data:

• Day-of-week structure: often the first clear signal, and often not the day people expected.
• Sleep coupling: whether short nights precede low days, and with what lag.
• Social load: for some people low mood follows isolation; for others it follows too much company. Both are common and people are frequently wrong about which they are.
• The recovery curve: how long a bad stretch actually lasts. Reliably shorter than it feels from inside one, and knowing your own number is genuinely useful the next time.

Practical rules that help it survive

• Never backfill more than a day. A guessed entry is worse than a missing one; it pollutes the record you're relying on.
• Missing days are fine. Gaps are data too, and a broken streak is not a reason to abandon the whole thing. As with any habit, the rule that matters is never miss twice.
• Don't grade yourself. The record is an instrument, not a report card. A month of 3s is information, not a failure.
• Review monthly, not daily. Daily review invites reading noise as signal. Once a month is where the patterns live.
• Keep it private. Honest entries require somewhere you're not performing. This is a real argument for a journal that stays on your own device rather than syncing to someone's server.

Start smaller than you want to

The version of this that works is almost disappointingly modest: one tap for how the day went, a tag if something stands out, a sentence when there's something to say. It looks too simple to be worth doing, which is exactly why it survives long enough to become worth doing.

The elaborate version, the spreadsheet with correlations and the nine metrics, is more interesting to design and you will not be doing it in six weeks. Build the version you'll still have in a year, then look at what it tells you.