Ken Rudin makes a compelling case for on-demand BI in DM Review. If anything, he over-eggs things by saying that users will get more engaged with on-demand versions of software, and how you don’t really need IT deployment skills, which seems less than obvious to me. However the thrust of the argument is still valid. I know having been on both the end-user side of the fence (at Shell) and as a vendor (Kalido) that a scary proportion of problems that users encounter are “environmental” i.e. some combination of software installed on the PC causes an application to break, and the software support desk has great difficulty in replicating the problem because it is next to impossible to ensure that your PC has precisely the same patch release level of operating system/database/middleware as the customer who has the problem. Indeed this is why large companies go to great lengths to try to standardise the desktop configuration across the enterprise, doing upgrades as rarely as possible. This is no mean project if you have tens of thousands of desktops in lots of countries, and tends to results in customer frustration as they find that the latest feature their children are playing with at home is not available on their locked down and fairly stable but quite out of date software they have at work.
The lack of environmental intrusion seems to me the key advantage of the on-demand model to the customer. As a side benefit, but not really a function of the model, most vendors have changed their pricing models for on-demand so that customers pay on a leasing basis rather than a big up-front license charge. This has potentially benefits to both customers and vendors, since customers get to pay only on usage, which seems to them a fairer way of paying that having to pay up-front for something they are not sure about, and for vendors it gives them a steadier revenue stream, as well as a possibility to reduce the sales cycle: USD 100 per month per user sounds a lot less than USD 100k plus 20% maintenance, though of course it may not actually be if the usage rates become high enough. For example Actuate cite this as a reason why they are bullish on their on-demand offering, as they can employ relatively cheap “inside reps” (essentially up-market telesales people) rather than costly enterprise software salesmen, who may be able to land that big deal but so often do not in practice, yet still expect a hefty salary whether they sell anything or not.
It is indeed rather curious, then, as to why on-demand has not really taken off more in the BI sector. There are some toe-in-the-water efforts from some vendors, but I suspect that most are nervous that such sales cannibalise their conventional channels. To me it seems there are opportunities here for start-up companies who donâ€™t have entrenched ways of doing business to take advantage of this situation, though since the VC community is out frantically hunting for the next Myspace and Facebook it is scarcely funding anything so untrendy as enterprise software, so there is actually little activity here either. At some point the pendulum will surely swing back, and at this point companies who have embraced on-demand delivery seem to me to be well-placed to take advantage of a rare but genuine shift in delivery model.
I’d be interested in any BI vendors or end-users with experience of on-demand BI who’d like to share their experiences, good or bad.