This is fascinating. One question I always have with large personal datasets like this: at what point did you start getting genuinely surprising insights versus confirming things you already suspected?

380k datapoints sounds incredible but I imagine the real challenge is turning that into decisions that actually change behavior