I think if you tell people that denotational semantics is just model theory for programming languages you've got most of the way there.
Another consequence of this perspective is that you must care about nonstandard models, even if you think you don't! When you prove something by natural number induction, you are precisely constructing a non-standard model of Nat in Prop.