Sally could have had a child, but didn't. Does that mean that there is some possible child, of Sally's, who doesn't actually exist? Can we give him or her a name - 'Kim', say - and lament for Kim's own sake that s/he wasn't brought into existence?
I think such talk is metaphysically confused. But it can sometimes be useful to talk of mere possibilia (e.g. possible people), so long as we take care to understand what is really being said. Suppose that, if Sally had had a child, it would have had a wonderful, flourishing life. This modal fact might give us a reason to prefer that Sally had the child. We might express this in shorthand by saying, "Kim's welfare gives us reason to want Kim to exist". But we are not really talking about Sally's child Kim - there is no such person to talk about. Rather, we are talking about the world and how it could have turned out. It could have turned out that Sally had a child with a flourishing life, and this fact about the world gives us a reason to wish things had turned out that way.
Compare our talk of fictional characters: it is convenient to reify them, and talk "about" Frodo and Sauron and the rest. We can even say true things using such talk: e.g. Frodo destroyed Sauron's ring of power. This is a truth, not about what actually happened, but about the fiction. There is not really an entity, 'Frodo', and another, 'Sauron', such that the former destroyed the latter's ring of power. You can't reify intentional objects so. There aren't really any such people or things. But it is sometimes convenient to talk as if there were, since that can help us talk about particular features of the fiction.
In the same way, we can talk "about" possible people in this loose, 'de dicto' sense. But it's just words. It would be a mistake to reify them, or to think that there are some particular possible people (de re) of which we speak. Talk of possible people should instead be understood as shorthand for talking about particular features of a way the world could be -- i.e. such that more people exist than currently do.
This theoretical point has at least two important applications: (i) in understanding what's wrong with the ontological argument, and (ii) in seeing that possible people cannot really be the ultimate source of the (even non-instrumental) reasons to bring them into existence.* More on this in a future post...
* = [Of course, there isn't really any 'them' to refer to. So that should read: "reasons to bring it about that the world contains additional people."]