What the research actually says about building a habit, in plain words and with the studies named. Each one opens with a direct answer to the question in its title.
A median of 66 days in the one field study that measured it, and a range from 18 to 254. Where the 21-day figure came from.
Every well-evidenced number next to the paper it came from, and the widely quoted ones that have no study behind them.
The novelty runs out before the habit takes over. Three plan faults explain almost all of it.
Small enough to do on the worst day of your month. Why that is the mechanism rather than the compromise.
Naming when and where you will act roughly doubles the odds of doing it. How to pick one that survives a bad week.
One missed day has no measurable effect. What happens on the second day is what decides the outcome.
Not wanting to is weather. Do the smallest version and decide about the rest afterwards.
One. The case is arithmetic rather than moral, and the numbers are worse than most people expect.
Smaller, at the same moment, tomorrow. A gap removes repetitions and cancels none of the ones you already had.
Shift work, travel and small children break the usual advice. Anchor to something portable instead.
The hour is the least transferable part. Build the structure out of one habit at a time.
Record daily, review weekly. Collapsing the two loses the only information worth having.
A habit runs without a decision; a routine is a sequence you decide to run. They fail differently.