Things that really convince me are comparisons with already existing stuff. Show what the problems are with an example program like c++ or java and show how d resolves the issue. Be careful though, one of the first things that put me off is when I feel it's too biased and they only show possitive points. "if something is too good to be true it usually is"