Cloudonomics: The Business Value of Cloud Computing, + Website

Paying on Actuals versus Forecasts

One benefit of SaaS in its current formulation is that, typically, you pay based on the actual number of subscribers rather than a projection. Often projected use of a new application is lower than expected: Users can’t seem to “get” the new system; they are too busy to attend scheduled training; other workarounds do the job just fine; access, security, language, or other implementation issues prevent certain functional groups, global regions, or groups with given devices or operating systems from being onboarded on schedule.

For all these reasons, paying as you go based on the number of users who actually are actively able to use the system or at least interested enough to acquire an account can cost less than an enterprise license sized to a higher number of users.

However, this benefit is not an insurmountable barrier for a legacy software vendor to overcome. There is no reason that licensed software couldn’t be priced based on the actual number of users or even finer-grained metrics, such as CPU minutes via one-minute term licenses or percentage of CPU utilization. This is not hypothetical: I negotiated such a deal with a major software vendor a few years ago.

The challenge to existing software vendors is not that they can’t do this, it’s that they’d rather not: They have a track record of financials based on collecting revenues on a front-loaded basis. A package is sold at a given price, with maintenance of, say, 20% to 25% per year. This ...