In Serre's Linear Representations of Finite Groups, there's a simple yet useful statement
(Section 8.1 Corollary to Prop 24): Let $G$ be a finite group and $A$ an abelian normal subgroup. Then any finite dimensional complex irreducible representation $\rho$ has dimension dividing $(G:A)$.
To me, this result is very satisfying (because it reduces the complexity of the spectrum of $G$ a lot if $G$ is not far away from being abelian, in which case the spectrum is simplest!), so I would like to understand it more. Unfortunately, in the book it's proved by induction. I hope to see if there's a more natural or constructive proof. In particular, I wonder if it's possible to naturally associate $\rho$ to a subgroup of $G/A$, with matching dimension and order.