How to read a month of habit data

The month ends. You look at the grid, notice it is mostly green, feel briefly pleased, and close the tab. Next month you do the same thing. Nothing changes, because looking is not reading.

A month of tracking is useful data, but only if you ask it specific questions. This is the review that takes about ten minutes and changes what you do in September.

Start with the rate, not the total

The first number to look at is completion against goal, per habit. Not the raw count. Twenty-two completions means nothing until you know whether you were aiming for 20 or for 30.

There is a second rate worth knowing, and the difference between them is where most of the useful information lives. Completion measures you against the target you set. Consistency measures you against the days that have actually happened. In a finished month they are close. Mid-month they diverge sharply, which is why judging a month in progress by completion alone will always make you feel behind.

Then read the shape, not the count

Two habits can both land at 70 percent and mean completely different things. This is the part a summary number cannot tell you and the grid can.

Scattered gaps, one here and one there, are just life. A habit at 70 percent with evenly spread misses is stable and probably does not need changing, only a slightly more honest goal.

Clustered gaps are different. A solid first half and an empty second half means something changed partway through the month, and the habit did not survive the change. That is worth naming: a work deadline, a trip, a change in when you wake up. The fix is almost always to attach the habit to something more stable, not to try harder next month.

A block of four or more empty days is the strongest signal in the whole grid. It rarely means you got lazy for exactly four days. It usually means the conditions the habit depended on went away.

Check the day of the week

This is the single highest-value thing in a monthly review and almost nobody does it by hand, because counting it manually is tedious enough that you will not.

Most people have one weekday that is quietly terrible. It is usually the day with a recurring meeting, the day of a long commute, or the first day of the weekend when the whole structure of the day changes. You will feel like you miss randomly. The breakdown will show you it is Saturday, every time, by a wide margin.

Once you know which day it is, the fix is concrete. Move the habit earlier on that day, shrink it on that day, or accept it and set your monthly goal to a number that has that day removed from it. All three are better than continuing to be surprised every week.

Look at perfect days separately

A perfect day is one where every active habit got ticked. It is worth tracking on its own because it measures something the individual rates do not: whether your habits fit together in a single day.

You can have four habits at 80 percent each and almost no perfect days, if they compete for the same evening hour. That is a scheduling problem hiding inside four perfectly respectable numbers. Few or no perfect days across a month of otherwise good rates is a sign to move one habit to the morning.

If you logged sleep and mood

The two-line chart is the part of a habit tracker most likely to be over-read, so a warning first. A month of one person's data cannot establish that anything causes anything. Sleep and mood move together for plenty of reasons, and both of them move with things you are not tracking.

What it is good for is spotting a shape you would otherwise miss. A run of five low-sleep nights that lines up exactly with the block of empty boxes in your running row is a useful thing to notice, not because it proves anything, but because it gives you a specific hypothesis to test next month. That is the right use of it: generating a question, not answering one.

The chart is for noticing. The next month is for testing.

Make at most two changes

The failure mode of a good review is leaving it with nine adjustments. Then next month tells you nothing, because you changed too many variables to know which one worked.

Pick a maximum of two. In rough order of what tends to pay off:

1. Fix a goal that was never honest. If you hit 21 out of 30 twice running, the goal is 21.
2. Move one habit to a different time of day, ideally attached to something that already happens reliably.
3. Cut or shrink the worst performer. Shrinking usually beats cutting: twenty pages becomes five.
4. Add a new habit, but only if an existing one passed a full month above 80 percent and survived a bad day.

Notice that adding is last. Most months should not include a new habit. A review that ends with fewer, better specified rows is a more successful review than one that ends with more.

The short version

• Read completion against goal per habit, and treat under 40 percent as a decision to make, not a number to feel bad about.
• Read the shape of the gaps. Scattered misses are fine, a block of four means conditions changed.
• Check which weekday is dragging you down. It is almost always one specific day.
• Count perfect days. Good individual rates with no perfect days is a scheduling problem.
• Use the sleep and mood lines to generate a question for next month, never to conclude anything this month.
• Leave with two changes at most, and usually no new habits.

Ten minutes at the end of the month, and the difference is that September is a response to August rather than a repeat of it.