Erik Keller has
an interesting editorial over on Sandhill.com, where he argues that there is a shift going on in corporate IT in favor of building more applications in-house (or through contract developers) instead of buying software packages. Keller made some of the same arguments a couple of years ago, and I
commented on them back then.
Basically, his argument is that there are three trends at work today that are making custom development a more viable option: service-oriented architectures (SOA), which make it easier to integrate custom software components into existing systems; the availability of open source, which can be used as a starting point for custom systems; and offshore development firms that are developing high quality code at low cost.
These points are generally well understood, although the implication that they represent a threat to software package vendors is not fully recognized. Keller made these same points two years ago.
But in his editorial this time, Keller points out something new: three negative factors on the side of commercial software providers that are fueling the trend toward custom development. Ironically, he says, these factors are exactly the same as those that worked in
favor of packaged software in the past.
According to Keller, these factors are (quoting him directly, here):
Slow time to market: Like the mainframe-oriented IT shops of 1980s, many of the largest enterprise-software vendors find it difficult and expensive to quickly incorporate the latest technology into their products in a timely and innovative fashion. They also tend to have limited experience with the latest tool sets.
Poor quality: Internal IT groups often used to fail when they attempted large, complex projects. Over the last 15 years, buyers have found that most enterprise software vendors and systems integrators are no better and actually less accountable than their internal capabilities.
High expense: With large upfront charges and on-going maintenance fees hovering around 20 percent of list price, enterprise software has taken on the same bad characteristics of inefficiently managed internal IT staffs.
As I wrote two years ago, I still believe that commercial software packages are the best route for most companies, especially small and mid-size organizations that are ill-equipped to maintain, let alone develop, comprehensive enterprise applications. For example, if a manufacturing firm has difficulty implementing SAP, or Oracle, or Great Plains, just wait until it attempts to develop time-phase material requirements planning or available-to-promise logic from scratch.
That being said, however, the economics of the build-vs-buy decision are difficult to argue with. By paying approximately 20% of the license fee for maintenance each year for Oracle or SAP, you are essentially buying the software again every five years. Although companies should continue to use commercial software for basic horizontal functions, such as finance and accounting, manufacturing, and basic order processing, the build-option is becoming much more attractive for complementary and industry-specific or company-specific functionality.
The transition of the major vendors toward service-oriented architectures (SOA) is making the build-option easier, with the ability to plug in such niche-functionality more easily. As Andy Bartels at Forrester pointed out recently, by embracing SOA, vendors such as SAP and Oracle are unleashing forces they cannot control, as the same SOA that makes it easier for vendors and partners to build composite applications also make it easier for customers to build their own composite applications.
The vendors' embrace of SOA is actually sowing the seeds of their own destruction. But they have no choice. They can either get on the SOA train or get run over it.
Related postsBuild/buy pendulum swinging back toward build
Send tips, rumors, gossip, and feedback to Frank Scavo at
. I'm especially interested in hearing about best practices, lessons learned, horror stories, and case studies of success or failure.
