Feature Importance
A ranking of how much each feature drives predictions, computed free from split counts or, more honestly, from the drop in held-out score after shuffling it.
Mean decrease in impurity, the version every tree library prints for free, is biased toward high-cardinality columns simply because they offer more candidate cut points to win a split, and it splits credit between correlated features rather than crediting either fully.
Permutation importance answers a more honest question: shuffle one feature on held-out data and measure how far the score falls. Neither version implies causation — a feature can be genuinely predictive and still not be something anyone can act on.