Build versus buy: resist the seduction.

Are you a software company? No? Then why are you spending millions of dollars a year building software? My software company uses photocopiers, but it doesn't manufacture them. We travel all over the country, but we don't build airplanes -- or pilot them.

For years, companies have spent huge sums of money developing software they believed would fit their organizations. That worked when companies operated according to the old, "vertical integration" model in which an organization owned every capability -- including sales, manufacturing, accounting, and management information services -- it thought necessary to manufacture and sell its product or service.

Today, that system does not work. Most companies now are trying to identify their core competencies. If you think one of them is software development, think again.

As part of the old, vertical integration model, MIS was a centralized factory, taking in requests for information and churning out data considered critical to running the enterprise. However, two funny things happened while so many companies were spending so many billions of dollars to design applications software. First, some of them realized they could sell the software they developed, or sell usage of it by smaller, similar companies. Second, new kinds of companies -- software companies -- emerged to develop packaged software for accounts payable, telemarketing, and other applications.

This new system worked fairly well. Packaged software, in particular, gave large organizations the ability to pay someone else to develop the software, so they could focus on their core competencies.

Unfortunately, the large IS empires that arose in the last few years are difficult to topple, Today, many companies continue to believe that maintaining sizable in-house development organizations to create applications is still the best way to go.

A recent cover story in an IS trade publication told of a grocery store chain that had fallen in love with a vision of itself as an object-oriented information superpower. An object-oriented approach provides a framework for integrating new applications with your existing applications. Not only had this company decided to re-develop all its software systems using objects, but it made the fatal mistake of trying to develop its own object technology and development tools to carry out this vision. The result: The company recently decided to abandon the object-technology initiative (sacrificing millions of dollars in research and development) and focus on what it does best -- "selling groceries," as the new CIO put it.

A 1994 survey of IS project managers, conducted by San Ramon, CA-based Center for Project Management, shows that over half of all IS projects come in between 60 percent and 190 percent over budget, and have less than 70 percent of the functionality they promised to deliver Worse, only one-quarter of IS projects are completed on time.

I'm not suggesting you trash your entire software-development organization. Your company certainly can benefit from on-staff software experts, and from re-training its designers and developers to use new client/server development tools.

However, I do recommend taking a long, hard look at your company to decide whether you really want to be in the software development business if it is not one of your core competencies.

So where do you go from there? Here are some options:

* Focus on user-enabling tools and applications. Your developers should find ways to mine the massive quantities of legacy data in your organization, and make it available to employees and managers. For example, using Windows tools such as Visual BASICS and a network connection to your mainframe, your developers can give end users a new perspective on your corporate data.

* Buy highly customizable, packaged software. Let the software company bear the overhead and the sales and marketing costs. Let its software experts stay ahead of the technology power curve and consistently enhance the package to better meet your needs.

* Outsource applications development. If you must develop your own applications, let someone else do the development and project management. You won't have to maintain a large, permanent staff of software programmers (a big IS cost).

The key is to decide what you do best. Most likely, you will discover it is not software development. If it were, your company would be a multi-billion dollar hospitality chain -- and you'd be as wealthy as Bill Gates.