Do the smallest possible version today, and do not try to make up what you missed.
That is most of the answer, and the two halves matter in that order. The smallest version (two minutes, one page, one set) is what re-establishes the cue. Making up the missed days is what turns a three-day gap into a quit, because the debt is larger than the habit and it arrives on the day you have the least momentum.
The rest of this is about the part people get wrong before that: deciding whether you are resuming or restarting, and finding out what actually broke.
The lapse is not the problem
A single missed day has no measurable effect on habit formation. The research that produced the widely-quoted 66-day figure found occasional misses did not disrupt the automaticity curve at all. What disrupts it is consecutive misses, and what ends habits is not the missing but the response to it.
The mechanism is well documented and has a name: the abstinence violation effect. A lapse gets read as evidence about yourself rather than about a Tuesday, as in "I can't stick to anything", and once the identity is the thing in question, quitting resolves the discomfort faster than resuming does. The gap between one missed day and abandoning the habit entirely is almost always this paragraph, not a scheduling problem.
Which suggests the only rule worth being strict about: never miss twice. The first miss is noise. The second establishes a pattern, and the third makes absence the default.
Resume or restart?
They are different situations and they need different responses. The dividing line is roughly two weeks.
Under two weeks: resume. The cue is still intact. Do the habit today, at the same trigger, in the same place, at the same size. Nothing needs redesigning; you are picking up a thing that still exists.
Over a month: restart, and shrink it. The cue link has decayed and, more importantly, something changed in your life that broke it. Rebuild at a fraction of the old size. If it was thirty minutes, make it five, and expect a few weeks before it stops requiring a decision. Starting again at the old difficulty is the standard way to fail twice.
In between: resume at half. Same cue, smaller behaviour, grow it back over a fortnight.
The mistake in all three cases is the same: resuming at the size the habit had reached at its peak. That size was supported by momentum you no longer have.
Find out what actually broke
A habit that stops rarely stops because of willpower. Four causes cover most of it, and each has a different fix:
The cue disappeared. You changed jobs, moved, stopped commuting, the gym closed. The habit was anchored to something that no longer happens. This is by far the most common cause and the easiest to miss, because it doesn't feel like a reason; it feels like you stopped caring. Fix: find a new anchor in the new routine, not more resolve.
The habit was too big. It worked while motivation was high and became negotiable when motivation returned to normal. Fix: shrink it to the version you would do on your worst day, permanently.
Life got harder. Illness, a newborn, a deadline, a bereavement. The habit didn't fail; it was correctly deprioritised. Fix: nothing, until the circumstance changes. Resuming a habit during a crisis is a poor use of the attention the crisis needs.
You no longer want it. Sometimes the honest answer. A habit you set up for who you thought you should be will keep failing, and the fix is to drop it rather than to keep restarting it every January.
Working out which of the four applies takes five minutes and is the difference between a restart that holds and the fourth attempt at the same thing.
What the data tells you that memory won't
Memory of a habit's history is systematically wrong: recent misses feel like the whole story, and a good month two months ago is invisible. This is where a tracker earns its place, and it is a different job from motivating you.
Three things worth reading off the record:
- Which day of the week fails. A habit that dies on Fridays has a Friday problem, not a discipline problem, and the fix is usually a schedule change.
- How long the previous run lasted. Knowing the last attempt reached nineteen days changes what "I never stick to anything" means.
- Whether the lapse has a shape. A gradual thinning out over three weeks says the habit was too big. A hard stop on one date says the cue changed, and you can usually name what happened that week.
The yearly heatmap is the honest version of your history: which weekday fails, how long the previous runs were, and whether the habit thinned out or stopped dead. Check-ins export as CSV whenever you want to look at the pattern yourself.
What it doesDesign the restart so the next lapse survives itself
Three adjustments, before you begin again.
Set a floor version and write it down. The two-minute variant you do on the worst day: one page, one set, one sentence in the journal. A habit with a floor does not have a binary failure state, and most lapses are the result of the choice being between the full version and nothing.
Schedule the rest days deliberately. A habit set to four times a week is not broken on Tuesday. A great deal of streak fragility comes from claiming "every day" for something that was never a daily habit, and then having to break it every week.
Count completions in a window, not consecutive days. "22 of the last 30" survives a missed Thursday; "streak: 0" reports that nothing happened when in fact twenty-two things did. This is also the more accurate number, which is a better reason than the motivational one.
If reminders were carrying the habit and stopped working, the failure was probably in the cue rather than in you, and the quieter method covers what to replace them with.
Restart today, not Monday
Waiting for the start of the week, the month, or the year is the most reliable way to turn a two-day gap into a six-week one. The date has no properties. What it has is a licence to not do the thing until then, which is the same licence that produced the gap.
The version of this worth internalising: the restart is not an event. It is doing the smallest version once, today, at the usual cue. Everything else, the diagnosis, the resized habit, the schedule, can be worked out tomorrow, after the chain has one link in it again.