I am having some thoughts about charging models and doing some "free thinking".
In the past hosts charged based on disk and bandwidth usage.Because we now buy bandwidth by the mbit, the cost is pretty stable and the average use per customer is relatively low.Disk space is getting cheaper (but as we keep three backup copies of everything its 4x the cost you might thing).
So, this leaves three costs:
1. Processor usage
Processing power is a finite resource.I have strated using quad processor servers which have a lot more resilience to usage spikes and are less likley to crash but it still means that if we have a few heavy users on one machine we can put much fewer sites on that server.This increases the cost per account to us and this gets passd on evenly to all customers.It may be possible at some point to reilably measure processor usage and charge based on that.
2. Support costs
Customers who require a lot of support cost us so much money we donot make a profit from them and this has to be recovered from profit made from other customers.This is unfair to customers who are competent or who are willing to read the documentation.How about charging a low fee for hosting and a per ticket charge for support?That would mean that competent customers paid less and those who needed more support paid more.
3. Customer acquisition
In some cases we pay more to acquire a customer than we ever make from them, so is there a case for charging less for repeat business or word of mouth customers and substantially more for those who are recruited through advertising?
This would appear to be unfair, but so might charging per support ticket.
These sort of issues do need to be discussed in the industry as the current low level of pricing is leading to unreliablilty and corner cutting (when it gets to the £1.99/mo level), but the cvurrent pricing models just don't reflect the real costs.