Seriously. With the amount often billed by consultants, clients aren't going to balk at double, triple, or even 10x the hosting costs of something like GoDaddy. (assuming reliability is there)

Also, I'd play with offering "unlimited" plans vs "cheap to upgrade" plans. Maybe the market you are targeting (if its a technical market, like django hosting mentioned above) associated "unlimited" plans with bad service or other trade-offs. A/B test until you're blue in the face.