The weekly habit review that takes 5 minutes

Most people check their habit data every day and review it never. Both halves of that are wrong, and reversing them is the highest-return change available to anyone who already has a tracker running.

Daily checking is counterproductive. A single day's completion rate reflects whether a meeting ran late, not whether your system works. Checking it daily trains a monitoring behavior — watching your own performance for reassurance — which is a reasonable description of anxiety and converts a measurement tool into a mood input.

Reviewing is where the value is, and almost nobody does it, because nobody has a routine short enough to actually run.

Log daily, review fortnightly

The split that works: log every day in about ten seconds, review every two weeks in about five minutes.

Two weeks is the shortest window in which a pattern is real. One week is too noisy — a single holiday, illness or deadline dominates it. A month is long enough that you will have absorbed two weeks of a bad configuration before noticing.

Fortnightly is also short enough that each review covers a period you still remember, which matters when you are trying to explain a dip.

The five-minute routine

In order. The order matters, because each step narrows what the next one looks at.

1. The weekly series, not the average (60 seconds)

Pull up completion rate by week for the last six to eight weeks. You are looking at two things and neither is the level.

Direction. 62% rising from 50% is a system coming together. 62% falling from 80% is a system coming apart. Identical number, opposite situations, opposite responses.

Volatility. A steady 70% every week is a stable system slightly below capacity. Alternating 95% and 45% weeks averaging 70% is something else — usually a novelty cycle or an unsustainable load.

Do not look at the monthly average. Averaging over a long window destroys exactly the information you came for. What your completion rate actually means covers the four ways the number misleads.

2. The category breakdown (60 seconds)

Which areas of life are carrying, and which are not.

The number to find: any category more than about 25 points below your overall rate. That is the thing to work on, and it is invisible in the aggregate.

This step also reframes bad fortnights usefully. A 38% period looks like collapse as one number. Across five categories it often shows Vitality at 80% and Order at 10% — which is triage, not failure. You protected what mattered. Seeing that is the difference between "I failed" and "I prioritised," and the distinction is real rather than consoling.

The pattern to watch for across several reviews: a category that is systematically pale every time. That is standing neglect rather than a bad fortnight, and it is the most common thing people discover in their first few reviews.

3. The worst single habit (60 seconds)

Overall rates are almost always dominated by one or two outliers. One habit at 20% can pull a whole fortnight down several points.

Find it, then pick one of four fates:

• →Fix the cadence. If it is set to daily and your life supports twice a week, that is the whole problem. This is the most common fix and the most frequently missed.
• →Shrink the behavior. A forty-minute version at 20% is worth less than a ten-minute version at 80%.
• →Re-anchor it. If the cue is unreliable, the habit inherits that. Habit stacking with ADHD covers anchor selection.
• →Drop it. If you would feel relief at removing it, remove it.

4. The weekday view (60 seconds)

The most underrated view in habit analytics. Most people have one structurally bad day and almost nobody knows which one until they look.

If one weekday runs 20 or more points below the others consistently, that is a scheduling problem with a scheduling solution. Move the habit off that day, or reduce what is scheduled on it. Attempting to solve a structural Wednesday with resolve produces a reliable weekly failure for months.

5. Change exactly one thing (60 seconds)

The discipline that makes the whole routine work.

Pick the single highest-value change from steps 1–4 and make it. Write down what you changed and what you expect to happen.

Then stop. If you change four things, the next review cannot tell you which one worked, and you will have learned nothing from two weeks of data. One change per fortnight is twenty-six experiments a year, which is plenty.

What not to do in a review

Do not add habits. Reviews generate enthusiasm, and enthusiasm added to a list produces over-subscription. Adding is a separate decision requiring a separate test: has the current set run above 80% for four consecutive weeks? If not, there is no spare capacity. How many habits should you track at once covers the ceiling.

Do not rebuild the system. The urge to restructure arrives when the data looks bad, and restructuring supplies a fresh hit of novelty that feels like progress. It is the most common way people lose their history. Rotating habits as novelty fades covers the controlled alternative.

Do not journal. A review is a decision procedure, not a reflection exercise. If reflection is valuable to you, it is a separate activity on a separate schedule — bundling it here is how a five-minute routine becomes a thirty-minute one you stop doing.

Do not interpret a single bad fortnight. If something structural happened — illness, travel, a crisis — the correct reading is that the instrument worked. No change needed.

When the review says do nothing

Worth stating explicitly, because a review that always produces a change is manufacturing work.

Three situations where the correct output is no change.

The numbers are stable and in range. If you have run 70–85% for several fortnights with no category badly lagging, the system works. The temptation is to optimise anyway, because you set aside five minutes and doing nothing feels like waste. Resist it — a stable configuration is the goal, and the main threat to it is your own restlessness.

Something external explains the dip. Illness, travel, a crisis, a bereavement. The instrument reported accurately and the configuration did not cause it. Changing the system in response to a one-off means running a crisis-tuned setup after the crisis ends.

You changed something last fortnight and the result is not in yet. Each change needs a full fortnight to produce readable data. If the picture is ambiguous, wait another fortnight rather than changing something else. Stacking changes is how you end up unable to attribute any outcome to any cause.

A useful practice is to write down "no change, stable" when that is the finding. It looks like an empty entry and it is the record that stops you fiddling — six consecutive no-change reviews is strong evidence you have found a configuration that fits your life, which was the entire objective.

The quarterly version

Every three months, a longer version worth about twenty minutes:

• →Re-weight. Anything running above 90% for two months that no longer takes deliberate effort should drop a weight level. A 4 that became a 2 is a success, and recording it keeps your percentages honest. Weighted habit tracking covers calibration.
• →Rotate what has gone stale. One habit, swapped for a new one in the same category so the trend line stays continuous.
• →Look at the full heatmap. Three months of grid shows seasonal structure that a fortnightly view cannot. Reading your habit heatmap covers the patterns.
• →Ask what the cumulative total says. The rolling average tells you about now. The permanent total tells you what you have built, and it is the thing that makes a bad quarter legible as a part rather than a whole.

Where to keep the notes

A practical question that determines whether the routine survives: the record of what you changed needs to live somewhere you will see it at the next review.

Inside the tracker is ideal if it supports a note. Otherwise, a single running document — one line per review, dated — is enough. The format that works is deliberately minimal: the date, the one change, and what you expected.

What it is for is preventing the most common failure of self-directed reviews, which is quietly making the same change repeatedly. Without a record, you will lower the same cadence three times across six months and never notice that it did not help the first two times. With one, the pattern is obvious in ten seconds.

Why this is worth doing at all

A tracker that you only log into is a diary. A tracker you review is an instrument, and the difference is whether anything changes as a result.

The specific value is that it moves your decisions off how last fortnight felt and onto what actually happened. Those diverge more than people expect — a fortnight that felt terrible frequently shows 70% with one bad category, and a fortnight that felt fine frequently shows the standing neglect you have been avoiding noticing for two months.

Five minutes every two weeks, one change each time. That is the entire method, and it is more than most people ever do with their habit data.

Further reading: the complete guide to gamified habit tracking, what your completion rate actually means, and habit tracking burnout for what happens when checking replaces reviewing.