What Habits Should I Track? What to Stop Tracking

Track a habit only while the record still changes a decision. In practice that means three things have to be true at once: the behavior is one you control, you would genuinely misremember whether you did it, and you would do something different this week based on what the log says. A habit that fails any of the three is charging you a daily logging tax and paying nothing back. The useful answer to "what habits should I track" is almost always a shorter list than the one you have.

That is not the answer this question usually gets. Search it and you will find posts offering 35, 55, 100 or 155 things to track, sorted into health, home, money, productivity and relationships. Those lists are answering a different question, which is what could be tracked, and the honest answer to that one is nearly any repeated behavior. Volume is not the constraint. Attention is. Every row you add to a grid is a small daily obligation, and the grid does not tell you which rows are earning it.

What habits should I track? Run the three-part test

Run each candidate habit through this before it gets a row. The point is to disqualify things, not to find more of them.

Condition	It passes when	It fails when
You control the behavior	"Walk 20 minutes" is a thing you either did or did not do	"Sleep 8 hours" or "feel calm" is an outcome that answers to things you cannot check off
Memory is actually unreliable	You honestly could not say whether you did it on Tuesday	You would know without looking, because it happens at a fixed time you never miss
The record would change something	A month of this data would make you move the time, cut the size, or drop the habit	You would look at it, feel something about it, and change nothing

Most abandoned trackers fail the third condition, and it is the one nobody tests for. Data you will not act on is not feedback. It is a diary of your own compliance, and keeping it feels like progress while producing none.

What is genuinely worth tracking

A habit in its first weeks. This is where the log does the most work, because nothing is settled yet and you are still deciding the time, the size and the anchor. The record is what tells you the 6 a.m. version was never realistic.

A habit you have restarted more than once. Repeat failures usually have a pattern, and the pattern is almost never visible from memory. It is visible in a log that records the days around the collapse.

Misses, not just completions. Most tracking records only successes, which means the interesting half of the data is missing. If you log the misses too, you can compute how fast you come back, which is the argument behind your return rate as a better number than your longest streak.

The trigger, when you are subtracting rather than adding. If the goal is to stop doing something, logging whether you resisted tells you very little. Logging when and where the urge fired tells you what to redesign, which is the whole method in breaking a bad habit by changing the trigger.

One habit at a time, in most cases. There is no researched number for how many habits a person can build at once, and anyone quoting you one is inventing it. The reasoning behind keeping it small is in how many habits you can honestly build at once.

What to stop tracking

Anything that has gone automatic. Once a behavior runs on its own cue without a decision, the log has no job left. Tracking it produces a satisfying row of checkmarks and zero information. The signs that a habit has crossed that line are set out in the real test for when a habit becomes automatic, and when it has, retire the row. A tracker is scaffolding, not architecture.

Outcomes you do not control. Weight, mood, hours slept, page views, income. These belong to a different category of record entirely. They move for reasons that have nothing to do with what you did that day, so a bad number reads as a personal failure when it is often just noise.

Anything you have never once acted on. Look back at the last two months. Which rows produced a change? The rows that produced nothing were never measurements, they were decoration.

The second and third metric on the same habit. Duration, distance, quality rating, mood before and after. Each extra column multiplies the logging cost and rarely changes the decision. One column per habit is usually the whole useful signal.

Rows you added because the grid looked empty. A grid with three rows looks sparse and a grid with twelve looks serious. That is an aesthetic preference wearing the costume of a system.

Mood, sleep and anything clinical are a different object

If the reason you want to track is that your mood is low, your sleep is broken or your anxiety is running your week, a habit grid is the wrong instrument and this site will not pretend otherwise. It has no diagnostic value, and treating a self-made chart as evidence about your mental or physical health is a category error, not a discipline problem.

A record of how you have been sleeping or feeling can be genuinely useful, but its use is as something you bring to a doctor or a licensed clinician who can interpret it. It is not something to grade yourself against. ProvenHabits writes about behavior design for people who are basically well. Where the question is clinical, it belongs with a professional, and no habit framework here is offered as treatment or prevention for anything.

What the evidence supports, and exactly where it stops

There is real evidence that monitoring helps. Benjamin Harkin and colleagues published a meta-analysis in Psychological Bulletin in 2016 (volume 142, issue 2, pages 198 to 229) covering 138 experiments and 19,951 participants. It found that prompting people to monitor progress toward a goal increased the chance they attained it, and that the effect was larger when the outcome was physically recorded and when progress was reported to someone else.

That is a strong finding and it is worth knowing. Here is what it does not say. It does not say that tracking twelve behaviors works better than tracking one. It does not say how long to keep tracking. It does not tell you which habits deserve a row. And it does not address the case where the logging itself becomes a burden, which is the failure mode most people searching this question are actually living inside. The study supports monitoring. It does not support a bigger grid, and it is regularly cited as if it did.

Two figures this article will not give you

A day count for how long to track before stopping. The popular 21-day number traces to Maxwell Maltz, a cosmetic surgeon, who wrote in Psycho-Cybernetics (1960) that his patients seemed to need a minimum of about 21 days to get used to an altered appearance. That is an observation about adjusting to a new face, not a habit-formation study, and it was never presented as one.

The 66-day number is real research but is almost always misquoted. Lally and colleagues, published in the European Journal of Social Psychology in 2010, followed 96 people performing a chosen daily behavior over a 12-week window. Sixty-six days was a median, and it was a median among only the participants who reached the automaticity threshold at all. Roughly half did not reach it inside the study window, and the range across those who did ran from about 18 to 254 days. Lally has publicly said the real finding is how variable habit formation is. So there is no honest "track it for N days then stop." The test is whether the behavior still needs a decision, not what the calendar says.

A maximum number of habits. No study establishes one. The percentages that circulate on this topic, the kind that claim a rule improves retention by some precise share, generally have no named study behind them at all, and you will not find any of them here.

When the number quietly becomes the goal

Charles Goodhart, writing about UK monetary policy in 1975, observed that any statistical regularity tends to collapse once pressure is placed on it for control purposes. The anthropologist Marilyn Strathern restated it in a 1997 paper on auditing in British universities, in European Review, as the line most people know: when a measure becomes a target, it ceases to be a good measure.

Habit tracking is an unusually clean demonstration. The measure is the checkmark. Once the checkmark is the point, a one-minute token workout is worth exactly as much as a real one, and it is cheaper, so that is what you start doing. The habit hollows out while the chart improves. This is also the mechanism behind why streaks end up punishing the people using them: the number becomes something you can fail at, in addition to the habit.

The defense is not a better chart. It is being willing to log an honest zero, and to stop looking at a metric that has started running you.

A ten-minute subtraction pass

Take your current tracker and go row by row.

1. Automatic? Cross it out. It graduated.
2. Outcome rather than behavior? Cross it out, or rewrite it as the behavior underneath it. "Lose weight" becomes "walk 20 minutes," or it goes.
3. Acted on in the last two months? If nothing changed because of this row, cross it out.
4. Would you notice its absence? If deleting the row would not change your week at all, it was decoration.
5. Of what survives, which one habit is the least settled? That is the one still earning a log.

Whatever is left is your tracker. For most people that is one row, sometimes two, and a note of the misses. If your grid drops from twelve rows to two, nothing has been lost except the work of maintaining ten rows that were not telling you anything.

The part nobody selling a tracker mentions

Tracking has costs, and they are real rather than theoretical. There is the daily time, small but compounding. There is the guilt of blanks, where a half-empty grid reads as a verdict on you rather than on the grid's design. There is false precision, where a month of checkmarks feels like evidence about a life when it is evidence about a chart. And there is the identity cost, where you become a person who tracks habits rather than a person who has them, and the maintenance of the system substitutes for the thing the system was for.

None of that means tracking is a bad idea. It means the friction has to be paid for by information, and the number of habits that can pay for it is small. The right method matters too, and the trade-offs between paper, a wall calendar, a spreadsheet and an app are covered in choosing a tracking method you will actually keep.

I abandoned nineteen habits over two years before any of this made sense to me. The correction was not a better system for recording them. It was recording fewer of them, and being honest about what the record was for.

Frequently asked questions

What habits should I track first?
One. Pick the habit that is least settled, meaning the one you have most recently started or restarted, and track only that. A single habit tracked honestly gives you usable information about your own week. Ten tracked carelessly gives you a wall of blanks and a reason to quit the whole system.

Should I track a habit forever?
No. The log's job ends when the behavior no longer needs a decision. Once it runs on its own cue and you would do it whether or not anyone recorded it, retire the row. Keeping it is work with no return, and it crowds out attention for a habit that still needs the scaffolding.

Is it better to track the behavior or the result?
The behavior, in nearly every case. You control whether you walked for 20 minutes. You do not control what the scale says on any given morning, because the result answers to things you cannot check off. Track the action and let the outcome be a slower, separate observation.

How many habits is too many to track?
There is no researched number and any specific figure you are quoted is invented. The practical signal is the grid itself: if most of it is blank most days, or if updating it feels like a chore you are behind on, it is too many. Cut rows until the tracker is something you finish in seconds.

Does habit tracking actually work?
Monitoring progress toward a goal has real support. Harkin and colleagues, in Psychological Bulletin (2016), found across 138 experiments that prompting progress monitoring improved goal attainment, with larger effects when the outcome was physically recorded. What that evidence does not establish is that more tracking is better, or that a twelve-row grid beats a one-row one.