the presentation looks confusing, but i think he means the graphs to convey that per card price was lowest with vistaprint
Right, I am puzzled too, especially since that according to his graph it was actually overnight prints that was the cheapest with better quality than vistaprint. I don't see how with these results the author can recommend vistaprint unless there is a mistake somewhere.
It looks to me like it's Moo for quality and speed, Overnight for price, and everything else is significantly worse in every dimension.