If a one hour single-interviewer interview is a significant improvement then it seems like companies should do this themselves, I'm not sure what service you are really providing unless you are just a really good interviewer.
I don't see how this is really true. You say an interview can use up 15-20 developer hours and in some percentage of the failure cases you are saving that (and hopefully not failing some percentage of the passing casses, btw.)
My point is that if a 1-hour screening by a single developer is effective, I can keep it in-house and save 14-19 developer hours where you'd be saving me 15-20.
I guess I am actually assuming you are only spending 1 hour with the candidate, you never say that above. But my point is, unless you are uniquely good at interviewing, the amount of time you are putting into interviewing the candidate is the amount of the you are saving me.