How to Engineer a Better Day, According to Data

Day Made: You know how some days just feel off and other days feel like you’d want to relive them time and time again? That gut feeling may actually have some math behind it. A study published last month in PNAS Nexus used machine learning to analyze American Time Use Survey data from over 15,000 people across 2013 and 2021, identifying which activities (and how much of them) predicted rating a day “better than typical.”

How to Do It: The researchers found that between 30 minutes and two hours of socializing was associated with people reporting having a good day. So was up to four hours of exercise, and five to six hours of hanging out with family and friends. No amount of housework or relaxing was linked to a better day, and more than two hours of socializing stopped adding benefit.

The Benefits: The models were about 62 to 63% accurate (better than chance but not quite 100% certainty). The real takeaway from the study is that variety wins. Dividing your day across connection, movement, work you love, and time with your favorite people beats out a day consumed with any one activity. But putting a time limit on each activity also matters.