A couple suggestions on your pricing... and this is just from my perspective, so take it for what it's worth.
#1, I think you should offer it for free for some number of rigs - some things, like SimpleMining do this, where it's free for 2 rigs or whatever, but then costs beyond that. I think it's appealing because it means you can test it without any real cost and for most of the single-miner people, they aren't going to be looking for a solution like this anyway.
#2, I would offer both a percentage and a purchase option - fees work fine for small to mid sized, but for a larger miner at some point it's just cheaper to hire someone to dev a similar project than pay the fees, so you miss out on them as customers.
Some general things:
#1, Your model for power price calculation is cool, but too simple to really to good cost estimation... I would recommend checking out the one in:
http://www.theenergydetective.com/We've used this at 3 mines each with different power companies/rate plans, and it works great. It's much more complicated to support, but if you're trying to use cost as a switching criteria, then you should try to calculate it correctly. I might have missed it, but it would also be nice to enter in the rig overhead - for example, with fans, etc a rig might have a base overhead of 100w, it would be nice to include that. This would also let you forecast what your power bill is going to be - we do that at all our places, and the TED is usually within a couple bucks of actual, which is awesome.
It also really needs a super simple dashboard view - something that can just be glanced at and see if anything is wrong. This could be as easy as just something that show a couple of green boxes - one for hash rate, one for rejected shares, one for rig count, one for gpu count, one for hardware errors. With each, just define what makes it change to color - for instance, it might be if it's 5% below typical hash it turns yellow, 10% it turns red. One machine under configured it turns yellow, two machines it turns red. It's also helpful to have total GPU's tracked, as occasionally you'll have a rig which will continue to mine, but one of the GPU's has gone offline on it - so this would catch that. And monitoring the reported fan speed would also let you know whether or not a GPU fan has failed - which is the most common issue we run into (although the Linux version doesn't report fan speed correctly yet, I've reported it and they say it will be fixed in an upcoming release).
Overall I think what you've done is cool - not quite cool enough for us to deploy for some of the reasons above, but cool enough to definitely keep an eye on and re-evaluate again later...
