It measures the effect of an infinitesimal contamination at a point on the estimator.

It is used to identify outliers in the dataset.

It represents the derivative of the estimator with respect to the empirical distribution.

It is always a linear function.