Habits take longer than you think

A commonly quoted figure is that it takes 21 days to form a habit. But there’s little meaningful evidence for this. More robust data comes from Phillippa Lally at University College London. In 2009, she recruited 82 participants and asked them to start a new habit. These were simple behaviours, such as drinking a glass of water with lunch or doing a press-up after brushing their teeth. On average it took 66 days until those behaviours were completed unthinkingly, which was Lally’s definition of a habit. But that single figure masked sizeable variation - 95% of people formed a habit in somewhere between 18 and 254 days. If you’re trying to reshape behaviour, don’t rely on a short burst of activity. You need more sustained interventions.

Twenty-one days is a magnificent commercial promise and a rotten description of habit formation. It offers a finish line, a certificate and just enough optimism to fit on a challenge page.

The study in the excerpt found an average nearer 66 days, with individual estimates ranging from 18 to 254. That spread is the important bit and the bit slogans murder first. “About two months on average, except sometimes much sooner or vastly later” has insufficient motivational sparkle.

The variation is hardly mysterious. Drinking water after breakfast, running five kilometres and practising a difficult skill place different demands on a person. Stable cues differ. Existing routines differ. Illness, travel, weekends and domestic chaos do not respect programme calendars.

One participant can feel automatic on day 18. Another can still rely on an alarm after day 100. Both facts can belong to the same evidence without either person representing a moral victory or failure.

The twenty-one-day myth turns ordinary variation into personal deficiency. Miss the ceremonial finish and apparently discipline failed. Complete the streak and struggle on day 22, and the certificate starts to look sarcastic.

At follow-up, the programme sheet offers three statuses: complete, incomplete, dropped out. Participants report changed cues, interrupted routines, automatic actions and inconsistent continuation. The administrator adds a fourth column because the original categories cannot hold the answers.