How long a golf swing change takes to stick

Where the 21 days came from

66 days, 18 to 254Median and range to 95% automaticity, Lally et al.

Three weeks is the figure a golfer hears most often, and nobody arrived at it by watching a swing. Gardner, Lally and Wardle, in the British Journal of General Practice, say it appears to originate in anecdotal evidence about plastic surgery patients who adjusted to a new appearance within 21 days. Their citation is a 1960 book, Psycho-Cybernetics.

What replaced it came from Lally and colleagues, who had volunteers repeat one chosen daily behaviour and rate how automatic it felt. The median time to 95% automaticity was 66 days, and the spread ran from 18 days to 254. The general practice paper turns that into advice: expect around ten weeks.

A 2024 review by Singh, Murphy, Maher and Smith pooled 20 studies and 2,601 participants. Median times ran from 59 to 66 days, means from 106 to 154, individuals from 4 days to 335. Every behaviour in that literature is about the size of drinking a glass of water after breakfast, not a movement rehearsed at speed with a club in your hands.

Ten thousand hours was never the finding

7,410 hoursPractice by age 18, best young violinists, Ericsson 1993

The second number arrives whenever somebody says a change takes years. Ericsson, Krampe and Tesch-Römer studied violinists and reported practice accumulated by age 18: 7,410 hours on average for the best young players, 5,301 for the good ones, 3,420 for those training to teach. Their abstract puts the requirement in years, a minimum of ten.

In the paper itself, 10,000 hours appears once, describing what more than a decade of deliberate practice adds up to. Ericsson and Harwell recorded in 2019 that Gladwell's Outliers gave a chapter to a ten-thousand-hour rule and cited the 1993 study as its evidence. Their verdict on the rule attributed to them: no evidence supported a magical number.

Neither figure transfers to your swing. One counts days of repeating a simple action in a fixed setting; the other counts a childhood spent building a skill from nothing. Altering one joint angle inside a movement you already own is a third thing, and nobody has put a stopwatch on it.

The end of a session measures very little

Seven days laterWhen the watch-only group's swing scores finally moved

Motor learning separates what you can do now from what you have kept. Ranganathan, Lee and Krishnan rank that learning-performance distinction among the field's most important insights: changes during practice are not always evidence of learning. So studies stop trusting the last block of trials and add a delayed retention test.

The interval matters. Their guidance is that a 24-hour retention test is defensible for a task practised across one day, and that longer timescales of learning need longer gaps. A swing change worked at across a season sits at the long end of that, and Sunday's verdict on Saturday's work is no verdict.

D'Innocenzo and colleagues had 21 novices learn a golf swing, seven of them simply shown a model swing to watch. Two PGA-qualified coaches scored every swing. That group was no better at the end of the session than before it started. Seven days later the same seven had gone from 61.1 to 71.1, and the difference reached significance.

The drop is part of the process, not a warning

Carson and Collins set out technical refinement in an already-skilled athlete as five stages: analysis, awareness, adjustment, re-automation and assurance. The second is deautomation, taking a movement that currently runs without you and putting it back under conscious control. Nothing gets rebuilt while it is still running on its own.

So their model does not treat the dip as an accident. Once the decision to change is taken, a fall in performance is inevitable as the process begins, and a player who keeps competing through it may pay in the result too: a worse ranking, a missed halfway cut.

The timescale they attach is up to 12 months, depending on the sport's complexity and how far the new movement sits from the old. Golf sits at their demanding end, because a golfer carries many skills rather than one. Read it as a practitioner's working estimate, not a measured average. The failure they name is stopping inside the dip: Woods, Harrington and Perks appear as golfers whose changes came apart under pressure.

What to count instead of days

If the calendar is the wrong unit, repetition in a stable setting is a better one. The habit work finds early repetitions add more automaticity than later ones, so the first fortnight of daily rehearsal is worth more than the second. Singh's review also reports that self-chosen behaviours formed more reliably than assigned ones, which argues for changing whatever actually bothers you.

Judge on a gap, not on a session. Film the same way each time and compare the frame you care about across clips shot on different days. Pebble Swing reads joint angles and the backswing-to-downswing tempo ratio out of the frames, which makes that comparison countable. What it cannot answer is whether the movement has gone back under automatic control: a frame shows a position, not who is steering it.