Is it okay to ask for help in clarifying a thought? Here's the particular thought I want help clarifying.
It seems that conservative extensions can alter the notion of structure homomorphism (but not isomorphism). In particular, adding new relations, even when they can be defined in terms of the old relations, may result in a stricter notion of homomorphism.
e.g. Consider a poset $(X,\leq)$ that is a meet-semilattice. We can conservatively extend this poset to produce a new structure $(X,\leq,\wedge)$ by defining that $x \wedge y = z$ if and only if $z$ is the least upper bound of $\{x,y\}$. This is extension by definition.
Now every (structure) homomorphism $(X,\leq,\wedge) \rightarrow (Y,\leq,\wedge)$ is necessarily a structure homomorphism $(X,\leq) \rightarrow (Y,\leq)$. But the converse may fail. That's because, in more cause speak, "A subposet of a lattice that just happens itself to be a lattice may fail to be a sublattice."