On a recent flight to Phoenix, I sat next to a veteran of a four-year implementation process for a client/server policy administration system. He and his new bride were heading for Sedona and some time off. Hed been doing 60-hour weeks for years and the software package his company had licensed from a major vendor was almost operational. He estimated theyd spent about $30 million for a regional implementation of personal lines. It almost killed the carrier and it had virtually no Web element. They had installed an obsolete solution at great cost.

I commiserated with this devoted IT professional. I too had been there, done that years ago. It was an awful process and one that, when all was said and done, wasnt worth it. In my case, my team had created software from scratch. But this carrier had bought what was supposed to be a pre-existing solution. How could it take so long and cost so much misery? Youve heard these stories before and perhaps even lived through them. Package solutions sometimes arent what theyre cracked up to be. And, by the time you tune the package to your unique requirements its an open question as to whether it would have made more sense to develop what you wanted right from the get-go.

But DIY system development is problematic as well. Its difficult to find and keep talent. Projects have a tendency to run over budget. Management imposes changes along the way. Scope creeps. Good project management intentions give way to crisis management. Its not a pretty picture.