The blogosphere isn't exactly aflame conversing about a canonical definition for software-as-a-service, but some good debate has been sparked. David Terrar wraps up the discussion with contributions from Thomas Otter, Nick Carr, Dennis Howlett and Vinnie Mirchandani, and then presents five questions as an SaaS litmus test:
1. Is the software behind the service properly web architected?
2. Is the solution using a multi-tenanted, 1 to many model?
3. Do they use true “pay as you go” pricing?
4. Do they have a “self service” component in their approach to support?
5. Do they have a SaaS business model?
David fleshes out the questions and there is some good discussion following his post.
Update: ZDNet blogger and SaaS explorer Phil Wainewright adds his insights to the discussion on his blog:
What really matters is delivering business services, not software. People go to ADP not because they want computerized payroll — if that's what they want they can buy a software package to do that — but because what they really want is to get their payroll done on time. They don't care how ADP achieves that goal. They just want a service performed, accurately, reliably and cost-effectively. Of course, to achieve those objectives, any service provider these days is going to have an automated infrastructure that uses some pretty powerful software. It so happens that ADP in the US uses SAP. But the software is not the service. It's just part of the infrastructure.
The other technology piece that any service provider is increasingly going to be using today is the Web — 1.0, 2.0 or whatever x.0 level of sophistication is available and appropriate. And it's this combination of the Web with software infrastructure powering on-demand business services that makes the killer difference.