Keep it at $0. That's the game.

• 281 adults had a goal of 7,000 steps a day. One group got $42 at the start of each month and lost $1.40 for every day they missed. They hit the goal on 45% of days [3].
• The same $1.40 paid as a reward for each day hit: 35% of days. A daily lottery: 36%. No money: 30% [3].
• People weigh a loss more heavily than a gain of the same size. In the words of the 1979 paper that set out prospect theory, losses loom larger than gains [1].
• Across 150 studies and 607 estimates, a loss weighed about twice as much as a gain of the same size [2].

What they don't show: The step trial was one employer, for 13 weeks, and once the money stopped, steps fell back to the same as no money at all [3]. How strongly a loss is felt varies a lot from study to study [2].

• People with a bias toward the present who don't see it in themselves put off tasks whose costs come now and whose rewards come later [4].
• Across 21 studies and 3,913 people, those with ADHD discounted money they'd get later more steeply than those without [5].
• Beeminder, which has run commitment contracts since 2011, says its stakes work by bringing long-term consequences near [6].

What they don't show: These come from a model and from money choices in studies. Neither tested exercise [4] [5].

• In a megastudy of 61,293 gym members that tested 54 four-week ideas side by side, the one that did best gave a small reward for coming back after a missed workout [12].
• Asked to put $150 down up front, 13.7% of smokers took part. Offered a reward, 90% did [11].
• Among people who would take either, the deposit helped more of them quit [11].
• Seeing a broken streak made people less likely to keep going. Being able to repair it softened that [13].

What they don't show: These were about smoking, gyms and other habits [11] [12] [13]. No study has tested a monthly price that moves with your weeks, as steps.org's does.

• Across six studies, people preferred goals with a few emergency reserves, and stuck with them longer [14].
• Some activity is better than none. Adults who aren't there yet should start small, then build up how often, how long and how hard they move [16].
• For reference, the WHO's guidelines for adults are 150 to 300 minutes of moderate activity a week, or 75 to 150 of vigorous [16].

What they don't show: The slack studies were about other goals than exercise [14], and the guidelines are about health [16]. Pick the goal you'll keep, at any pace: steps.org counts the minutes you moved.

• Their own money on a goal of 1 lb a week: 14.0 lb lost in 16 weeks, against 3.9 lb without [7].
• Workers paid to use the company gym went more while they were paid, and a year later the effect was modest. Those also offered a contract with their own money on it were still going more a year after the payments ended [8].
• Smokers offered a savings account they'd lose if they failed a test: 11% took it. The group offered it was 3 points more likely to pass at 6 months, and still ahead in surprise tests at 12 [9].

What they don't show: People who choose a contract may differ from those who don't: at the gym, people already going were the likeliest to sign one [8]. A review of deposit trials for weight loss found a small-to-medium effect while they ran, with one outlier set aside, and none after [10].

• People were 8.7 lb down when their stake ended, and 1.2 lb down 36 weeks later [15].
• In the step trial, steps fell back once the money stopped [3].

What they don't show: Neither trial tested a stake that never ends [3] [15].