We play a game. There are ten outwardly identical boxes, containing pieces of paper with ink marks on them that are unambiguous inscriptions of non-indexical sentences in the predominant local language. Five of these sentences are true, and five are false. You get to pick out a box. If it's one of the ones containing a true sentence, you get $100, and if it's one of the ones containing a false sentence, you pay $100. Then, there is a fact of the matter whether a given box is such that were you to pick it, it would be correct to declare you the winner. But it is hard to avoid the conclusion that in some sense—perhaps not the ontologically beefiest if one is a sparse property theorist—five of the boxes have the property of being winning boxes and five of them have the property of being losing boxes. Moreover, these properties are not ontologically rock bottom. They plainly depend on geometrical properties of the ink marks within the boxes, and, further, on the truth of the sentence determined by these geometrical properties. It seems that if truth is not a property, then neither is being a winning box a property. But being a winning box is a property, and so is truth.
Anything that is a basis for objective classification is a property, and truth, plainly, is a basis for objective classification of boxes into ones that under the rules are winners and ones that under the rules are not winners.
Moreover, this example shows that truth is explanatory. That a given sentence is true can explain why you owe me $100. In fact, this case shows that just about any fact is explanatory, since one can center a game on just about any fact. This won't impress non-realists about normative states of affairs.