A fitness app paid users 4 cents a day. Two years on, they walked 242 more steps

Most evidence about fitness apps comes from trials lasting weeks. A study published this month in the British Journal of Sports Medicine follows 516,818 users of a commercial app for two years, which makes it one of the few datasets capable of answering whether anything sticks [s1].

What was studied

The analysis is a 24-month pre-post quasi-experiment conducted in Ontario, Canada, running from December 2016 to June 2019 [s1]. The app was a multicomponent commercial product with a distinctive feature: after a one-to-two-week baseline period, users earned very small financial incentives — $0.04 CAD per day — for hitting daily step goals [s1].

Four cents a day is roughly $15 a year. The design question the study is really testing is whether an incentive that trivial can move behaviour at population scale, and for how long.

The sample was 62.83% female with a mean age of 33.46 years (SD 12.65), and 47.15% were classified as "low" active at baseline, meaning fewer than 5,000 daily steps [s1].

A quasi-experiment is not a randomised trial. There is no control group of comparable non-users, so the estimates are changes from each user's own baseline, and anything else that changed over those two years — seasons, life circumstances, the reasons someone downloaded the app in the first place — is not controlled away.

The average result, and why it is the least interesting one

Daily step counts were higher than baseline at all key timepoints. At 24 months, the average difference was 242 steps per day (p < 0.001) [s1].

That is a small number. The authors describe it as a modest increase of about 250 steps per day sustained over two years [s1]. Two hundred and forty-two steps is on the order of two minutes of walking.

The subgroups, which move in opposite directions

The population average conceals two much larger effects pulling against each other.

Among users classified as "low" active at baseline, the difference at 24 months was 1,986 steps per day (p < 0.001) [s1]. Among users who started earlier and had longer exposure to the incentives, it was 758 steps per day (p < 0.001) [s1]. The authors note that for these subgroups the increases approached or surpassed 1,000 steps per day, a level they characterise as indicative of clinical significance [s1].

Among users classified as "very high" active at baseline — 10,000 or more daily steps — the difference at 24 months was −3,969 steps per day (p < 0.001) [s1].

That last figure is substantial and the paper flags it directly [s1]. Its interpretation is genuinely uncertain. Regression to the mean is one candidate: people who joined during a peak activity period would be expected to drift down regardless. Seasonal enrolment effects are another, and the study's own finding that start season influenced outcomes [s1] is consistent with that. A behavioural explanation — that a micro-incentive tied to a step goal well below what a highly active person already does could anchor effort downward — is speculative and the study does not test it.

Whatever the cause, it means the headline average of 242 steps is a net of a large gain among inactive users and a large loss among very active ones, not a small gain shared by everyone.

A short randomised trial pointing the same way

A randomised controlled trial published the same day tested a related question over 12 weeks: whether app-delivered performance feedback beats health education alone [s2]. It enrolled 150 staff at Nnamdi Azikiwe University in Nigeria, mean age 47.11 ± 6.46 years, allocated to app-based performance feedback with hourly reminders (n = 52), the app without reminders (n = 50), or a control group (n = 48) [s2]. All three groups received education on cardiovascular risk factors and the benefits of physical activity [s2].

Both app groups showed significant improvements in pedometer step counts, Google Fit metrics, walking scores, moderate and vigorous physical activity, total activity and reduction in sedentary time (p < 0.05 across comparisons) [s2]. The version with hourly reminders outperformed both the app-only and control groups on pedometer step counts, and produced significantly greater sedentary-time reduction than control (F = 3.405, p = 0.038) [s2]. At baseline, 14% of participants had low activity levels, with no significant differences between groups [s2].

The two studies converge on an unglamorous conclusion: app-based prompting produces real but small effects, and the effects are larger in people who start with more room to improve.

Limits

The Ontario study observed an app that operated between 2016 and 2019 — a different product generation from what is on phones now. It measures steps, not health outcomes; no mortality, cardiovascular or metabolic endpoints are reported [s1]. The Nigerian trial ran 12 weeks and used a self-report questionnaire alongside its device measures [s2].

What to watch

Whether any commercial app publishes retention-adjusted long-term data. The 516,818 figure counts users in the dataset; how many were still opening the app at 24 months, and whether the step gains belong disproportionately to those who stayed, is the question that would settle how much of this generalises.

Sources

• Can fitness apps work long term? A 24-month quasiexperiment of 516 818 Canadian fitness app users — British Journal of Sports Medicine, 2025-09-23
• Mobile-application-based performance feedback improves physical activity behaviour compared to education alone: a randomised controlled trial — BMC Digital Health, 2025-09-23