To what extent could you eliminate on-site failure points and/or build-up on-site redundancy? The reason I ask is that maybe the question shouldn't be about hiring more people for on-site failures but instead planning for up-front costs to build-up redundancy at each location (ie RAID, virtualizing, better networking equipment, IP-KVM's, etc.).

I know local hardware failure will never be completely eliminated but your pitch could be "look, I'll provide your much better IT than you'll find locally at a much lower cost but to make this work we have to invest in your hardware to build-up redundancy and backup systems."