The eigenvalues of a unitary matrix lie on the unit circle. What are some applications in which the eigenvalue distribution of the matrix is important? For instance, that the eigenvalues are clustered, that they live in some half plane, etc. I'm particularly interested in distributional properties which is "global" in the sense moving a few eigenvalues (potentially arbitrarily far) or many eigenvalues a small distance doesn't really change the meaning. So for instance, that the matrix is in $SO(n)$ is not such a property because we could move a single eigenvalue from $-1$ to $1$ to change this property.
Likewise, an explication that the eigenvalues do not have meaning is also useful. For instance, I know that the unitary matrices from many factorization are only unique up to scaling by diagonal matrices with entries on the unit circle. Do such scaling preserve any meaningful properties of the spectrum?