Steve Mills runs IBMs $12 billion software business, which, for the most part, has weathered the technology downturn pretty well—so far. IBMs DB2 database and WebSphere application server have been strong performers this year, but the companys Tivoli Systems Inc. systems management business has fallen off a cliff. But Mills knows that from CIOs on down, IT managers care more about how software helps them solve problems than about a vendors financial numbers. So eWeek Editor at Large John Dodge, speaking to Mills in IBMs office complex in Somers, N.Y., last week, quizzed him on what makes IT projects succeed or fail.
eWeek: Why do many IT projects fail?
Mills: I always found it to be amazing that these things were being debated because there is not a company in the world over the last 20 years that could have the number of products, services, transactions, touch points in the marketplace or whatever your metric is without information technology.
eWeek: You dont believe the disaster rate was ever that high? Many surveys say the success rate is dismally low.
Mills: I dont think the disaster rate is high. Projects often ran too long or suffered from scope creep. That is rampant. Historically, IT projects tended to be fairly long projects—18 months, two years and three years. Today, we talk about projects of much shorter duration, with six-month milestones.
eWeek: Isnt integrating different platforms and systems still one of the biggest problems?
Mills: Part of the problem with software is you cant see it, smell it or taste it. You cant kick it, so you dont know how the thing is going to perform until it finally gets up and running.
eWeek: Doesnt that imply software projects are a crapshoot?
Mills: They are, to some extent. One of the toughest problems is cross- platform systems. They are tougher to deal with than mainframe environments. Mainframes are very manageable in contrast to a Unix environment or an Intel [Corp.]/Windows environment, where you dont have that integrated set of capabilities. This is where we run into the most problems.
eWeek: What is IBM doing to minimize the risk with software implementation?
Mills: We are very big on encouraging customers to do benchmarks of different kinds. I have people that spend all their time on that.
eWeek: Speaking of metrics, do you know what percentage of IT projects involving IBM software succeed?
Mills: I dont start any project that I dont believe I can be successful in.
eWeek: Is there a number relative to the ones that succeed?
Mills: The goal is 100 percent satisfaction. Where we run into problems is bringing the right resources to bear. Thats the thing that frustrates me the most. I have 25,000 technical people. I have 2,000 in-the-field technical practitioners. I have 4,000 field technical support reps.
eWeek: Youre the top software guy at IBM. Do you get called in when there is a big problem with a customer, and if so, how often?
Mills: Im usually in serious problem situations six to 10 times a year.
eWeek: Can you provide me with some specific examples?
Mills: Im not going to give you customer names, but Ill talk specifically of proactive health checking. We do an annual health check on Macys.com by looking at them in the fall, figuring out where the sensitivities and pinch points are. Macys runs very high traffic during the Christmas season, so you get them ready in August and September for an end of October, November and December push. We do the same thing for online brokerage firms with heavy transaction systems.
eWeek: Conversely, where is the low-hanging fruit in terms of payback?
Mills: Creating a new customer interactive presence that I can get up and running in six months. Renovating the supply chain with hundreds of major suppliers requires longer cycles. In those cases, you stage the projects so that they start to realize some return on investment [quickly] even when the complete project may take years.
eWeek: Does the success of an IT project depend more on company culture, or are there industries where its inherently tougher to succeed?
Mills: Financial services have a particular advantage because they are very volume and resource-reduction driven. Manufacturing companies have a tougher challenge because they are typically dispersed and highly decentralized.
eWeek: What impact is the economic downturn having on IT projects?
Mills: Its slowing the decision-making process. Of course, were used to outrageous growth. The industry this year is coming in at 8 to 9 percent growth. Customers will only spend a "trillion three" on information technology this year. I would hardly call that a bad business to be in.