The main point of cross validation is that it:

specifies where to

tells you to prune the tree

gives you an estimate of the performance of your trained model

calculates an impurity factor of your trained model when used on different data