What Should I Track About Myself? Start With One Number

2026-09-26 · Yesterself

Some complaints have no edges. Everything feels off. I'm tired all the time. I don't know where the money goes. Something about this year isn't working. You can feel the shape of it and nothing else, and eventually the thought arrives that you should probably start tracking something.

The instinct is sound. What usually follows it is not. What follows is a download: an app with forty fields, a template with mood and sleep and water and steps and gratitude, and one dutiful week of filling in every column. By day nine the entry costs four minutes. By day twelve it is skipped. And the nine days you did complete answer nothing, because nothing was ever asked.

The work is in three decisions made before day one. The logging is the easy part.

Write the complaint down in the words you actually used

Not the tidy version. The unfair, specific one you would say to a friend at the end of a bad week.

Everything's been off since about August and I can't tell if it's work or just me.

Now look at what is already nearly solid in that sentence. Since about August is a date. Work or just me is a fork with two ends. Those are the parts to keep. The vague word — off — is the part you must not try to measure yet.

Skip this step and you will end up tracking mood, because mood is what the apps offer, when the complaint you actually had was about your evenings.

Turn the complaint into a question with two possible answers

A complaint describes a state. A question has an answer that could come out either way, and that is the entire difference.

Three tests for a question worth a fortnight of your attention.

1. It contains a comparison or a time frame. "Am I tired?" is not a question; you already know. "Is it every day, or is it Wednesday to Friday?" is.
2. You cannot already write the answer. If you are confident how it will come out, you are not investigating, you are building a case.
3. Something would change. If both answers lead to the same week, this is trivia. Interesting, but it will not survive day fourteen.

Notice the shape of the good ones. Almost every useful starting question is is it A or B, not how much:

• Everything feels off → Is it every day, or only on the days I don't leave the house?
• I'm exhausted → Does it follow how late I went to bed, or what I did that day?
• There's nothing left at the end of the month → Is it a few large things, or many small repeated ones?
• I hate my job → Is it the whole week, or two particular hours in it?

An or question is far cheaper to answer than a how much question. How much needs precision and a scale you trust. A or B needs only enough resolution to separate two possibilities, which crude notes manage fine.

Write the question at the top of wherever the number is going to live. That one line is what keeps the log from becoming a habit with no purpose.

Pick the cheapest number that could answer it

Cheapest, not best. A crude measure you still have on day thirty beats a rigorous one you abandoned on day six, and it is not close.

A first metric should be:

• One field. Not three.
• Under twenty seconds. Timed, not estimated. If it needs an app to open, a sync to finish and a menu to choose from, it is over budget.
• Recoverable. If you forget until the next morning, you should still be able to fill it in honestly. A number destroyed by one missed evening will be destroyed.
• Already lying around, where possible. Your phone's own screen-time figure. The count of transactions on a statement. Whether the calendar had a meeting before ten. Data you don't have to generate costs nothing, and nothing is hard to beat.

Then one final check: if this comes out high, and then low, do I learn which side of my question is true? If not, it is the wrong number, however satisfying it would be to collect.

Worked through:

• Is it every day, or only on days I stay in? → Y or N for whether I went outside, plus a 1 to 5 for the day. Two characters and a digit.
• A few large purchases or many small ones? → The count of separate purchases each day. Not the amounts; the count answers the fork.
• The whole job, or two hours of it? → A 1 to 5 for the worst hour of the workday, and which hour it was.

That last one shows the general trick. When a complaint is vague, don't measure the average of anything. Measure the extreme, and note where it sat.

Set the stop date before day one

Write the end date beside the question, before you have logged a single thing.

This is the step people skip and it does the most work. A log with no end never produces a verdict; it accumulates, becomes an obligation, and eventually stands as evidence of your own inconsistency. A log with a date on it finishes.

How long: two weeks if the thing moves day to day, four if it moves week to week, one full cycle if it is tied to one. Longer is not more rigorous.

Two rules for the window.

Change nothing while it runs. Not bedtime, not caffeine, not the calendar. The urge to start fixing arrives around day four. Fix and measure at once and you will not know which one moved.

A missed day is a blank, not a failure. Leave the gap; don't fill it in from memory the next morning, because a remembered number tends to agree with the theory you already hold, and that theory is what you were testing. A broken run is not a reason to start over — chasing an unbroken streak quietly changes what you are willing to write down.

On the stop date, read the whole thing at once and write four sentences: the question, what the number did, which side of the fork it landed on, and what you'll do about it. Then actually stop. If the question turns out to have been the wrong one, you spent two weeks and bought a better one. That is the ordinary way this goes.

Keep one line of words beside the number

A bare number is almost unreadable six months later. A 2 in October means nothing on its own. A 2 beside third night up with the baby means everything.

So write a caption, not an entry. Ten words. It is the difference between data you can read later and a column of digits you will delete. If the number and the note disagree, the note is usually the more honest of the two. There is also the question of when you record it — a score written at midnight mostly describes the evening.

A small log is worth keeping somewhere it comes back to you. Yesterself is a private journal built on that return: an evening's writing arrives as a letter one morning later, and any sentence can be postmarked to a future date — re-read the fortnight and see whether it still holds, dated three months out. It is end-to-end encrypted, with no account and no password, and export is free forever. It isn't released yet; the waiting list on the site is the only door.

When to skip all of this

If the complaint is severe, persistent or frightening — weeks of it, or something that has stopped you functioning — the first step is a conversation with a doctor, not a spreadsheet. A tracker describes; it does not diagnose, treat or manage anything, and this is not medical advice. Bring the notes to the appointment — they make a short visit more useful. Just don't let collecting them become the reason to postpone it.

Common questions

What should I track about myself first?

One number, chosen to answer one question you have actually written down. The question comes first. Without it you are collecting data rather than answering anything, and a month of unfocused logging usually answers nothing.

How many things should I track at the start?

One. A second field roughly doubles what the day costs you and more than halves the odds you are still logging in week three. You can add more later, once the first has earned its place.

How long should I track before I look at the results?

Decide the end date before you start, and keep it short. Two weeks for anything that varies day to day, four if it varies by week. Longer is not more rigorous, it is only more chances to quit.

What if there is no pattern at all?

That is a real answer and worth writing down. It means the thing you suspected is not driving the complaint, which usually narrows the next question faster than a positive result would.

Should I use an app or paper for this?

Whichever you will still have open on day twelve. For a single number, the notes app you already use, the back of a diary, or one column in a spreadsheet all work equally well.