When modeling a shape that is going to be animated, it is important not only that you have enough control loops to shape the form of your object, but also that you don't have too many. You want to avoid situations where you have to excessive numbers of points just to define a new shape.  
 

It's best to learn to predict the behavior of your smoothing algorithm (whether nurbs, patch, bevel edge smooth, or subdivision surfaces). By being able to predict what will happen when you smooth an object, you'll make the most of the edges you use to define the shape of the object. Having a "lighter" model will also make other animation tasks much faster (animating, rendering, wiring). 

  Also, think about how the animation is going to show off your creature.  If your project has nothing but medium to faraway shots, don't focus so much on the hoof but more on the shape of the neck and back.  

Above all, spend the detail where it counts.  One trick I use to make sure I'm managing things well is to put the object in wireframe and zoom the camera out.  Whichever area of the wireframe model  
turns solid first is where you're spending your detail, and it should be the focus of your animation.