We investigate the use of a highly scalable intervention to help individuals achieve their physical activity goals: self-funded commitment contracts. Using a randomised experiment, we compare the effect of soft (non-incentivised) and hard (incentivised) contracts among 1629 members of a gym in Stockholm, Sweden. We find a significant positive impact of being offered a hard contract on monthly visits to the gym and the probability of meeting one's contract goal. Relative to the control group, being offered a hard contract significantly increases gym visits by 21%, while being offered a soft contract leads to an 8% increase.