Every product has a metric that predicts whether a new user will still be there in 90 days. Not the obvious one.
Facebook's was 7 friends added in 10 days. Twitter's was following 30 accounts in the first session.
Neither shows up in a standard dashboard. Both came from looking at what retained users actually did in week one - and tracing it back.
In crypto it might be the second trade. Or a recurring buy turned on. Or a card payment that felt indistinguishable from any other but cheaper.
The signal lives in your retained users. What did they do in week one that churned users didn't?