Here is the link for the article how Microsoft is dipping it's toes on how to improve cloud computing service level agreements. [link]

This describes how service levels need to be improved or even added as most software-as-a-service vendors do not provide SLA's at all not to mention report monitoring and availability metrics. Most interesting is the importance of violations of a service level and how to deal with incidents.

A refund of a service outage is the improvement and an addtional step from where we are today for software-as-a-service. In example, see a few sommers back with traditional electrical companies service outages in New York City. Cloud computing at Microsoft, Google and Amazon is internet centric computing simular to the traditional electric company i.e. PG&G. If an outage at PG&G occurs your are not billed for actual unsed service. Today in Software-as-a-Service your are billed monthly regardless if the service is available or if the service is being used at all.