Every January, a wave of people decide this is the year they finally learn something. New gym membership, new language app, new instrument. And every February, most of them quit.
I wanted to test a narrower version of that folk wisdom with real data: does the calendar month you start a habit actually predict whether you are still doing it six months later? Not "do resolutions fail" (they do), but "is the resolution window itself a bad time to start"?
There is a clean dataset for this. Skoove, an online piano-learning platform, tracked six-month retention by signup month across 1,137,446 learners from 2021 to 2024. The dataset is open on GitHub, so you can reproduce every number below.
The headline figure
Here is the finding in one sentence: the December-to-January resolution window, which pulls in roughly a quarter of all signups, has the worst six-month retention of any part of the year. People who start in spring stick with it far more.
