One fairly common objection to moral realism is that facts and properties can't be inherently prescriptive in the right kind of way. As Williams puts the non-cognitivist's argument (in 'Ought and Moral Obligation'), "if [a moral claim] just tells one a fact about the universe, one needs some further explanation of why [we] should take any notice of that particular fact." The objection may be pressed especially strongly against the non-naturalist, since it at least makes sense to care, however contingently, about natural properties like health and happiness. But, the critic may ask, why would we ever care about some non-natural property Q? On this way of looking at things, to identify reasons with some non-natural property Q would essentially undermine our practice of caring about reasons. One may make this vivid by way of an example: Suppose that gratuitous torture had the non-natural property Q that moral realists talk about. Would you then want to engage in gratuitous torture? Surely not...
However, I think this objection is badly misguided. Firstly, note that the normative supervenes on the non-normative, so gratuitous torture (i.e. where there are no instrumental benefits to so acting), if actually unjustified, could not possibly have a different normative status. Insofar as we consider it a priori that gratuitous torture is wrong, it's not clear that we can even make sense of the scenario we're being invited to consider here.
Secondly, a subtle point: To avoid moral fetishism, ordinary agents should not care directly about the property of rightness (or being a reason) itself; rather, they should care about the right-making features (or those natural properties which possess the non-natural property of being reason-giving). In other words: normative properties are higher-order properties. They tell us which first-order features we should care about. And it is those first order features, not the normative properties themselves, that are the proper objects of our moral concern. The significance of the normative properties is more indirect: they provide the standard for whether, in caring about and responding to some (natural) considerations and not others, we are caring about, and responding to, the right considerations.
With that clarification out of the way, we may still ask: why care about those natural features that happen to have this non-natural property Q? Isn't possession of a non-natural property just completely irrelevant to the importance of (say) relieving suffering?
Well, no. Of course, if all you know is that some feature has "some non-natural property Q" then you can't, from that uninformative description, tell the significance of this. So let me restate the previous questions in a more transparent way: "Why care about those natural features that have the property of being worth caring about? Isn't possession of the property of being worth caring about just completely irrelevant...?" When stated this way, the questions seem silly.
The critic may object that it's my use of the normative words 'worth caring about' (or 'being a reason', etc.) that's doing all the work here. And (they may add) that's cheating. If normative realism is true, then normativity should reside in the properties themselves, and not just our way of thinking and talking about them.
Fair enough. But it's a mistake to start from our uninformed grasp of some generically described "non-natural property", and from there interpret realists as saying that reasons are nothing over and above whatever uninteresting arbitrary property we previously picked out. It's not as though non-naturalist realists want to reduce normative properties to some independently identifiable non-natural property. (Open Question Arguments refute supernaturalist reductionism no less than naturalist reductionism.) Rather, the central claim of moral non-naturalists is that normative properties are distinct from all the properties we learn about elsewhere (cf. property dualism about qualia). It would go a bit too far to say that normative concepts are the only way to refer to normative properties -- obviously indirect discourse is possible, e.g. "the property that Richard last referred to" -- but we can at least say that the nature of normative properties can only be fully grasped via normative concepts. Even if one manages to refer to a normative property in some other, indirect way, one cannot grasp the nature of the property by such means. It remains opaque. (Again, much like property dualists hold of phenomenal concepts and phenomenal properties...)
This raises interesting questions about the nature of normative concepts, and how it is that they pick out the non-natural properties that they (according to realists) do. But that's a topic for another day. For now, I simply want to defend non-naturalist realism from the charge that the non-natural properties it posits cannot be normative. It's true that independently identifiable properties can't be inherently normative: our agreement on this point is precisely why we're not metaethical naturalists! But the critic has not yet done anything to show that there cannot be inherently normative properties that can only be transparently grasped by means of our normative concepts. And that's all that the non-naturalist is committed to here.