Let's say I have some nice mathematical construction $C$ that takes in (say) a group, and outputs (say) a topological space.
Well, groups don't exist in a vacuum! There are lots of homomorphisms between groups, and we learn a lot about a group by learning about the properties of the homomorphisms into and out of it. Similarly for topological spaces - with "homomorphism" replaced by "continuous map."
That is, both $\{$groups$\}$ and $\{$topological space$\}$ are not just sets (fine, classes), but categories. The functoriality of $C$ is the property that $C$ "plays nicely" with this categorial structure: roughly speaking, given a group homomorphism $f: A\rightarrow B$, I should get a continuous map $c_f: C(A)\rightarrow C(B)$ in some reasonable way. Informally, I want to say that $C$ "really is" a functor (although of course this is kind of an abuse of terminology; initially at least, $C$ might only be defined on objects).
Note that you could have a map on objects which can be "upgraded" to a functor in more than one natural way; however, in practice this doesn't tend to happen very often.