Effectiveness of Step Goal Personalization Strategies on Physical Activity in a Mobile Health App: A Field Study.

Users who set their own goals (personalized-by-you) increased weekly step count by 3793 steps compared to the not-changed group.

• 230 users chose to set their own goals in the personalized-by-you group
• Comparison was made against the not-changed group (933 users who chose to keep the default goal)
• Analysis used propensity score matching and difference-in-difference analysis
• The default goal was 2000 steps per day, 5 days a week

Users whose goals were set by the algorithm (personalized-by-the-algorithm) increased weekly step count by 4315 steps compared to the not-changed group.

• 236 users opted for algorithm-set goals
• The algorithm used a moving-window percentile rank method based on step data from the previous 4 weeks
• The two personalization strategies (personalized-by-you and personalized-by-the-algorithm) appeared to have a similar effect on step counts
• Analysis used propensity score matching and difference-in-difference analysis

Users in the not-changed group who kept the default goal also increased their weekly steps by 1759.

• The not-changed group consisted of 933 users who responded to the email but chose to keep the default goal
• This increase occurred without any goal personalization intervention
• This suggests a possible response or engagement effect from being contacted by email

The personalized-by-you strategy was effective for medium- and high-active users but not for low-active users.

• Medium-active users increased by 5842 steps per week
• High-active users increased by 4266 steps per week
• Low-active users increased by only 384 steps per week, which was not statistically significant (P=.82)

The personalized-by-the-algorithm strategy was effective for low- and medium-active users but not for high-active users.

• Low-active users increased by 5095 steps per week
• Medium-active users increased by 5278 steps per week
• High-active users increased by only 1446 steps per week, which was not statistically significant (P=.51)

The study involved 5800 users of SamenGezond, a Dutch mobile health app, divided into two random groups.

• Group 1 consisted of 3800 users who received an email asking about satisfaction with their current goal
• Group 2 consisted of 2000 users who did not receive the email and retained the default goal
• Of the 3800 users emailed, 1399 responded: 230 chose personalized-by-you, 236 chose personalized-by-the-algorithm, and 933 chose to keep the default goal
• All users had a default goal of 2000 steps per day, 5 days a week