The likelihood function can be factored into $g(T(x), \theta) h(x)$.

The likelihood function is independent of $\theta$ given $T(x)$.

The estimator $T(x)$ is always the Maximum Likelihood Estimator.

The variance of $T(x)$ reaches the Cramer-Rao Lower Bound.