Having witnessed the collapse of the first generation of ASPs due to the lack of a business model, my biggest bugbear about on-demand services and applications in the Web 2.0 era has always been, So how do we make money?
The answer to this question will determine the prospects for most of the wierd and wonderful Web 2.0 startups and wannabees parading their business concepts at last week's O'Reilly conference (described by a worried Phil Becker in his email newsletter last week as "the many young entrepreneurs running around the conference talking about the right exit structure before they know what business they are in (other than building that big online community to 'flip'.)"
The per-user, per-month model used today by most on-demand application vendors works well enough for discrete applications, but the more mash-ups of individual services there are, the more essential it will be to devise a successful model for billing services based on actual usage. As yet, no one has achieved that. So ever since Amazon announced last year that it would charge for use of its web services as soon as its Alexa Web Information Service came out of beta, I've been waiting with bated breath to see what it would do.
It's been a long wait, but finally the answer came today with the emergence of the production version of AWIS. Quoting from Amazon's press release, "Developers can use AWIS to answer difficult and interesting questions about the Web, and programmatically incorporate these answers directly into their applications." One enthusiastic user is Philip Kaplan's Adbrite, which "uses AWIS to provide prospective advertisement buyers with important information and statistics about the Web sites selling site space through the AdBrite marketplace."
Here's the charging model that applies to Adbrite and anyone else using AWIS as from today:
"There is no charge for the first 10,000 AWIS requests made by a developer each month and each additional request costs only $0.00015. For example, if a developer makes 12,000 AWIS requests in a month, the cost would be just $.30."
At 15c per thousand requests, a developer is either going to need to collect a lot of information or sustain a huge volume of traffic before this gets to be a serious amount of money. It would be fascinating to know the calculations underpinning this pricing model. Did Amazon do it on a cost-plus basis, or is it pricing to secure market share? What target customers did it use to model the ideal pricing? All these questions go to the heart of whether or not this approach to billing for on-demand services is going to be viable, and it's going to be fascinating to watch this experiment progress.