The reason it takes different lengths of time for different habits

Two people adopt two new routines on the same morning: a glass of water after breakfast, and fifty sit-ups before bed. Ten weeks later the first is automatic and the second still requires a decision every night. Both were performed with the same consistency. Only one has become a habit.

What everyone sees

Popular accounts supply a single number — twenty-one days, or sixty-six — and treat it as a property of habit formation in general. Failure to reach automaticity in that window is then read as a failure of discipline. The number is doing work it cannot support, because it was never a constant.

What is actually happening

Lally and colleagues tracked daily self-reports of automaticity and found a median of around sixty-six days with a range spanning eighteen to well over two hundred, varying systematically with the complexity of the behaviour. Wood and Neal’s framework explains the spread: automaticity develops through repeated cue-response pairing, and effortful multi-step behaviours accumulate that association far more slowly than single-step ones.

Why it stays hidden

The variance hides because a single number is more transmissible than a distribution. A median travels through articles and courses while the range is dropped, and the dropped range is exactly the part that would have made the slow case legible. The person whose behaviour sits at the long end therefore has no framework except personal failure.