Sizing the on-demand market and its growth potential
Research on emerging markets is notoriously unreliable — my advice to investors and entrepreneurs wanting to know the size of the opportunity in any new area is that, by the time we can reliably quantify the market, the best opportunities will already have gone.
The on-demand market is still very young, and therefore any attempt to size it is a developing science. But as I mentioned the other week, in my view any analysis of it ought to include the likes of Google — in which case, based on its quarterly figures announced yesterday, we're already looking at a multi-billion dollar market. Google's contextual advertising is clearly an on-demand service — buyers go online (or connect to an API) to place it and manage it — and last quarter it brought Google revenues of $1.58 billion. This is quite different from, say, Amazon.com, most of whose revenues come from the margin between its buying and selling prices, and which therefore don't count as services revenues. Google's service is entirely virtual, and entirely of its own creation (even when it serves ads to third-party content sites).
On reflection, though, I've come to the conclusion that Google's AdWords service isn't complete enough to fully qualify as an on-demand application. Although historically speaking the API came later than the online user interface, I think we have to categorize AdWords as an API-based service, and ignore its rudimentary web interface and tools as a bare-bones offering that it makes available for the convenience of less sophisticated users.
Anyone who's serious about using AdWords, along with any of its main competitors, will use a third-party application to manage their keyword buying and ad placement. WebSideStory's on-demand application Bid is an example of these much more sophisticated add-ons, using the API to get data and then presenting it in a way that aggregates and automates many of the reports and processes that a serious user requires.
What I'm suggesting, then, is that there are two tiers to the on-demand market. There's an on-demand services tier that consists of services designed for access via an API, and then there's an on-demand applications tier that provides automated data collation and process automation to users. While some applications are completely autonomous services, others exist to take one or more API-based services and layer automated processes on top of them. At the moment, the API-based services market, dominated by contextual advertising and transactional services such as eBay and Amazon Web Services, represent the largest market by far. But all the current excitement about mash-ups is probably an indication that ultimately the largest market (or perhaps just the one that makes the fattest profits) will be in the application layer that sits on top of the raw API services and assembles them together in innovatively productive ways.