15 statistics on habits, streaks and stakes

Every number below comes from a named study or dataset, listed at the bottom of the page. Where a figure is company-reported rather than peer-reviewed, it says so. If you use a number, take its caveat with it.

How long habits actually take

1. A new daily behaviour takes a median of 66 days to feel automatic, and the range across individuals ran from 18 to 254 days. The study tracked people for 84 days and fitted a curve to each one, so the top of that range is projected rather than observed. [1]

2. The famous 21 days was never a research finding. It comes from a 1960s plastic surgeon describing how long patients took to adjust to amputations, and it has been quoted out of context ever since. [2]

3. For gym habits specifically, tracking of new members found it took roughly 4 sessions a week for at least 6 weeks before the habit could hold itself up. Below that dose, it tended to collapse once the initial push ended. [3]

When people quit

4. About two thirds of exercise dropouts quit inside the first month. In supervised trials that followed 947 adults for six to eight months, most of those who left did so while still ramping up, and the reason they gave most often was lack of time. [4]

5. The people who survived that first month mostly stayed: adherence held above 80% for the rest of the programme. The hard part sits at the front, and it ends. [4] More on why that window is so lethal in the real reason you stop working out, and on getting through it in how to stay consistent when motivation dies.

What moves the needle

6. People whose favourite audiobooks only worked at the gym went up to 51% more often than controls. This is the original temptation bundling experiment. The 51% belongs to the group whose audiobooks were locked to the gym entirely; a looser version got 29%, and both are study averages, so individual results vary. [5]

7. When the restriction was later offered as a paid option, 61% of participants bought it. People don’t just tolerate constraints they chose; they pay for them. [5]

8. In the 2,000-person replication, the exercise boost was still detectable 17 weeks after the intervention ended. [6]

9. A conditional $80 reward, paid only if a gym-visit threshold was met, bought an average of 3.51 extra visits. [7]

10. Self-funded commitment contracts, where your own money is refunded only if you follow through, produced gym-attendance effects still detectable a year after the incentive ended. The group that only got paid incentives faded within months of the money stopping; the group that had staked its own money did not. [8]

11. stickK’s platform data across 600,000+ contracts claims financial stakes roughly triple the success rate, rising to 78% versus 35% when a referee verifies progress. Company-reported and not independently audited, so treat it as directional. [9]

What blocking and friction do

12. A pause before a chosen app opens, nothing more, cut how often people actually opened it by 57% over six weeks. Openings, not minutes, which is the honest reading of the number. [10]

13. In a hard digital-detox study, only about a quarter of participants hit the strict success bar, yet 91% improved on at least one measure of attention or wellbeing. Rigid targets get missed while real gains still happen. [11]

The honest ceiling

14. Paid incentives for health behaviours generally fade within about 3 months of the money stopping. Incentives rent the behaviour; contracts that cost you are the version with long-run evidence. [12]

15. When 54 interventions were tested on 61,293 gym members at once, 45% raised weekly gym visits while they were running, and only 8% still showed a measurable effect after the four weeks ended. Plenty of things work while you are doing them. Almost nothing kept working by itself. [13]

The pattern across all fifteen: plans and reminders are cheap and weak, stakes and bundling are the reliable levers, and everything works better when it is chosen in advance and verified by something other than your own report. The mechanics of that last part are in commitment devices explained.

Sources

1. Lally, P. et al. (2010). European Journal of Social Psychology, 40(6). University College London, 96 participants adopting a new daily behaviour, tracked for 84 days.
2. Maxwell Maltz, Psycho-Cybernetics (1960); the observation concerned patients adjusting to amputations.
3. Longitudinal tracking of new gym members; see also Buyalskaya et al. (2023) on when gym-going becomes context-insensitive.
4. STRRIDE randomised exercise trials, 947 adults, six-to-eight-month supervised programmes.
5. Milkman, K. L., Minson, J. A. & Volpp, K. G. M. (2014). Management Science, 60(2), 283-299. 226 participants.
6. Kirgios, E. L. et al. (2020). Organizational Behavior and Human Decision Processes, 161(S).
7. Carrera, M., Royer, H., Stehr, M., Sydnor, J. & Taubinsky, D. NBER working paper.
8. Royer, H., Stehr, M. & Sydnor, J. (2015). American Economic Journal: Applied Economics, 7(3).
9. stickK.com platform statistics, company-reported.
10. Grüning, D. J., Riedel, F. & Lorenz-Spreen, P. (2023). PNAS, 120(8). 280 participants over six weeks.
11. Georgetown University summary of the Kushlev digital-detox study, using the Freedom app.
12. Mantzari, E. et al. (2015). Preventive Medicine, 75.
13. Milkman, K. L. et al. (2021). Nature, 600. Megastudy: 54 interventions, 61,293 gym members, 30 scientists across 15 universities.