Enterprise Software in 2006 (and beyond) - Part 1

There's a lot of meat on Signal vs. Noise's New Year's prediction bone ("Enterprise is the new legacy"). Here's a few of the assumptions and statements:

1) Enterprise solutions cost 10x - 100x that of "simpler, lower cost solutions"
2) Enterprise vendors will begin hemorrhaging customers
3) People that use software [in companies] have different objectives than those who buy it
4) Enterprise vendors costs will continue to rise and quality will continue to drop
5) The term 'enterprise software' will leave the 2006 lexicon tarred with the same brush as 'legacy software' leaves 2005

You know, the thrust of their argument is correct, if not the specifics. Enterprise software is undergoing a real change right now, and in the near future the installation of business-facing, behind-the-firewall software will be much rarer than it is today. So, I'll use the SvN list to discuss the enterprise software market in more depth. Here's the first of 5 posts discussing their points in more detail...

1) Enterprise solutions cost 10x - 100x that of "simpler, lower cost solutions"

Look, we heard these exact same statements when we moved from mainframes to mini-computers. And then we heard them again when we moved to PC's... yet overall IT spend as a % of revenue has remained roughly constant since the '70's (in the F500/G3500). The issue of cost isn't the application, it hasn't been for many years. The issue is the cost of the data.

Of course what SvN is really saying is that we're on the move from software deployed behind the firewall to hosted solutions ala salesforce.com. This is partially true, but misses the mark on what's required to make this happen in a big way. The costliest part for IT and business today isn't the cost of developing some new business-facing workflow... it's integrating the many silos of data needed for the process! Unless we can radically change the cost of getting to data, we cannot radically alter the cost model of enterprise computing.

Salesforce.com has proven the hosted business model can be viable, but I think it's done something even more remarkable: it's gotten organizations over the hump of allowing a hosted service to hold very sensitive corporate transactional data, like a company's sales pipeline and revenue forecast. Today, via salesforce.com, many companies hand it over without a worry. This is a big shift, and paves the way for the really big news about SOA: exposing secure services via the net so that they can hook into hosted, business-facing solutions.

SOA-based standards biggest benefit isn't the "plug and play service orchestration" that some BPM advocates talk about, the real story is that services will expose enterprise data, securely, via standards-based protocols to "simpler, lower cost" business applications hosted anywhere - even outside the firewall.

This is why the rise in importance of architecture is so critical. To avoid the problems of the earlier distributed models, we need to make sure enterprise architectures better support the new distributed model. If we don't then we will, once again, end up with silo'd and redundant data everywhere. Your project management data will be on pm-host.com, some of your worklfows will be modeled bpm-host.com, and some of your word documents will be on Microsoft Live. And costs will not go down.

The initial moves to distributed computing - and some of the unachieved "benefits" we banked on - serve as cautionary tales, and CIOs are trying to apply those lessons learned. Architecture will be the foundation upon which we design our service-based data models, and the WS-* standards will be the doors (and deadbolts) through which we gain access.

The other thing we've learned is that, from a business perspective, the biggest benefit of salesforce.com isn't its lower cost, it is that the cost is paid directly for usage. There is a higher correlation of payments per seat vs. usage on salesforce.com, for example, than seats purchased to seats used for Siebel. This means that, as a business user, I have more direct correlation between what I buy and what I use, increasing productivity per dollar spent. It isn't "cheaper" most business people look for, it's "useful." By correlating payments and usage more directly, we'll see higher quality business applications.

What does this mean to BPM as we know it today? It changes it radically. The interesting, valuable, part of BPM as a technology will be more about enabling visibility into process than about automating it. Automation will happen by any number of means: hosted sites, custom enterprise apps, legacy apps. Just like today. Execution of your data will be increasingly complex, not increasingly simple. That's the nature of distributed systems.

So as we mature our business process management skills, process execution becomes the least valuable attribute of stand-alone BPM software; its ability to deliver process visibility and understanding becomes the more important aspect.

Tomorrow, I'll tackle how on Earth we're going to deal with "hemorrhaging customers."

Happy New Year!