Evan Schuman's StorefrontBacktalk

For years, one of the advantages of working in IT in a large chain has been the ability to deploy in-house and best-of-breed strategies as needed. It was that ability to craft a package that precisely fit the demands of a project that made some of those jobs enviable.

But as dollars have tightened, a funny thing happened: many of those packaged apps starting getting a lot better. And vendors including SAP and Oracle even tried crafting their on best-of-breed applications via unbridled acquisitions. (OK, many of them resembled Dr. Frankenstein’s best of breed, but it’s the thought that counts.)

New stats coming from research at RISNews.com are suggesting that, as suspected, homegrown packages are losing their home and roll-your-own are being rolled out the door. (No more. I promise.) IT execs saying that their “preferred IT deployment philosophy” is “deploying packaged applications with few modifications” came in at 53.8 percent, almost twice as popular an answer as “best of breed” at 28.8 percent. Even though RISNews has seen that trend for years, the difference had never been so stark. While the packaged response has been right around today’s 53.8 percent figure for years (ranging no farther than from the low- to mid-50s), best-of-breed has suddenly plummeted. It was at 46 percent in 2006, 45 percent in 2007 and 47 percent in 2008, before plunging to 28.8 percent this year, said RIS News Editor-in-Chief Joe Skorupa.

“CIOs have come to realize they do not want to be in the software platform development business. Today’s retail ERP applications are so complex that even Wal-Mart with thousands of developers is increasingly turning to packaged applications and suites. This problem is most acute with ERP apps, but it parallels many other application categories as well, including price optimization and supply chain execution,” Skorupa said. “These apps have matured and become proven to the point where retailers cannot build them as well at any price. Best of breed isn’t dead, because there are still many good packaged point solutions that are built on open standards that retailers can and should deploy in a best of breed way, say task management and planogramming, for example. But what is clearly dying is building home-grown applications. It will never die out completely to solve problems associated with small, niche services, but it is dying out very quickly for anything on the enterprise-class level.”

That’s all true, but it’s only part of the issue. The extraordinary resources needed to create and maintain those apps is simply too expensive today. Plus, CFOs are becoming more savvy to the proprietary lock-in element of homegrown or highly customized packages, where the programmers who created them become indispensible. It’s always convenient to let the creators of customization manage it, until they leave (or, much worse, threaten to leave) and then IT management has to pay the price for that homegrown convenience. Say what you will about Oracle and SAP, but they’ll always be there to pick your pocket. The “pick your pocket” part is frustrating, but the “will always be there” part is often ignored until you remember what the homegrown repairs involved.

Of course, vendors can go away, too. (There was a time that businesses assumed GM would be around forever. No one’s making that assumption today.) But at least they leave behind a massive installed base, who can either help each other or they’ll be worth enough money that some ISV will make it their business to take care of them as a group.

If the economic troubles go away in 2010 as some predict, will the roll your own numbers go back up? Not a chance, unless CIOs have very short memories. In other words, of course they will.