To ensure the probability of at least one false positive is less than $\alpha$.

To control the expected proportion of false discoveries among all rejected null hypotheses.

To maximize the power of all tests simultaneously.

To eliminate the need for $p$-value adjustments.