A couple of weeks ago I happened across this entry from Chris Jablonski discussing Gartner’s bearish forecast for Software-as-a-Service (SaaS) vendors; on demand services will account for less than 10 percent of the business applications market through 10%.  Their justification for this figure is:
The on-demand model is not suitable for complex business uses like logistics support and order handling, and for large complex companies requiring business process support.
Gartner apparently goes on to recommend avoiding SaaS when:
You are dealing with transactional-intensive applications such as in a warehouse management system; when data is exceptionally sensitive, and when on-demand service providers dont have the deep functionality or provide the level of customization required.
Interestingly, there have been some recent comments from notable SaaS advocates that align with that view. I was a bit surprised, for example, to see one of the more ardent advocates of the SaaS or as he might prefer it, the network computer model, Sun’s Jonathan Schwartz, quoted as saying the following:
Is AJAX or a browser an appropriate vehicle for heavyweight office productivity software? Absolutely not
Now it’s possible of course that the quote was taken out of context, or that there was more to Jonathan’s explanation. It’s also possible that that’s the only reasonable answer from a vendor that sells rich client office productivity software.
But I’m skeptical of all of the above. It’s not that I believe that Ajax is the ideal vehicle for office productivity features, or that SaaS is primed to solve complex transactional requirements. It’s rather that I so dramatically underestimated the potential for rich client applications a couple of years ago that I’m reluctant to do so again.
To explain, let’s go back to the year 2000. A younger, less jaded version of myself is slogging away as a systems integrator; building systems, managing projects, even (gasp) coding a bit here and there. Perhaps betraying the tendencies that led me to the analyst industry, I tried in my little spare time to keep up on industry trends and directions. One of those trends was that of the Application Service Provider (ASP) model, and I have to be honest – given my experiences, I found the very concept appalling. Not because it threatened the SI business – I was actually in favor of anything that let me not worry about the desktop, and I’d been around technology long enough even then to realize that people are unlikely to be factored out of the equation in my lifetime – but because looking at it from my customers point of view, I couldn’t fathom how an enterprise would ever agree to a scenario in which they:
- would pay money but own nothing
- wouldn’t be able to customize their application
- would ever outsource their critical data
- would force on their employees a terrible click/refresh, click/refresh model
This attitude was in spite of – or pehaps because of – the fact that one of the projects I worked on was the web enabling of a client server medical systems application. As far as I was concerned, the whole concept was a pipe dream, or a tiny niche play at best. And as it turned out, it mostly was. With very few exceptions, the folks making those “rent your applications” pitches are now gone. The larger folks like Microsoft obviously haven’t gone anywhere, but the startups that built their business that way largely dried up and blew away.
So in the context of that particular time, I was right to be skeptical. But in obsessing over the tactical shortcomings, I missed the longer term promise (it was before I became an analyst though, I promise – now I’m infallible . Put another way, I was focusing too much on what the technology couldn’t do, and not enough on what it could. Worse, I was implicitly assuming that that imbalance would be static.
Well, here we are five years later and I think we can all agree that things have changed a bit. My original concerns about the model? In light of the success of applications such as Gmail and Salesforce.com – and Larry Ellison saying that on demand is “vital”, they seem a bit quaint. Taking them point by point:
- Own software? who wants that? Would I rather pay a large up front fee for a single codebase that evolves at a glacial pace, or a far smaller regular fee for an application that can add new features over night?
- Customization? Go talk to large failed CRM or ERP implementations and see why they failed? 9 times out of 10 its because the customer got customization happy. Customizations introduce are a short term relief for users but a long term problem for IT because they’re simply not manageable. And if even knowing that you still want to customize, well some of the SaaS providers will accomodate you.
- Outsourcing critical data? Having worked at one shop that did CRM implementations, I can tell you for a long time that this was the showstopper for many enterprises; they might like the promise of SaaS, but they could not sanction under any circumstances the outsourcing of their customer data. That would never get beyond the firewalls. Well, what a difference a couple of years makes; some of those same customers are happily pushing customer back and forth to Salesforce.com.
- Terrible interface? Well, Ajax might not have solved all the browser based interface problems, but I think it’s increasingly clear that the gap between thick and thin clients is narrowing. Maybe Vista will reset that a bit, but we’ll have to see. Who would of thought five years ago that the browser would be not just a capable but honestly usable interface for email or CRM? Not I, certainly, and yet here we are. How bad can the browser really be if I spend upwards of 80% of my day in one?
Now some of you might be inclined to believe from the above that I’m a Software-as-a-Service zealot, ready to evangelize SaaS as the solution to every problem, but such is not the case. As I wrote a couple of weeks ago, I’m a believer that the desktop and its application set is going to be important for a long time to come. There’s ample room for both approaches.
But personally, I’m not inclined to put any application category – office productivity or otherwise – permanently out of the reach of thin clients, because they’ve already come far further than I or many others expected them to. Too often we’re condemning these technologies for what they can’t do today, without considering what they might be able to do tomorrow. Fool me once, shame on me, fool me twice, well, you know how it goes.
 Incidentally, does anyone else think that @ 10% of the entire business applications market is not so bad?