Three Mistakes HP Made, And One It Didn'tThree Mistakes HP Made, And One It Didn't

Every business technology team makes mistakes; it's just that most CIOs won't admit them. Here are three that Hewlett-Packard CIO Randy Mott and his team made during the company's three-year IT transformation--along with one they didn't.

Underestimated "Shadow IT": How bad did they miss? Going in, they thought the company had about 3,500 applications. It was more like 6,000. They knew they had 85 official data centers. Once they added them all up, they found more than 400 locations with servers or storage supporting applications. Most wouldn't qualify as data centers, but they had enough computing power that they needed to be moved to a data center and the applications had to be factored into the mix. The lesson: Take time to do a precise inventory before laying out an IT transformation plan.

Didn't Plan Well Enough For Acquisitions: It's hard to believe in retrospect, since acquisitions were part of HP's overall business transformation plan. Even before the EDS megadeal, HP did about 30 acquisitions the past three years, including one for Mercury that doubled the size of HP's software business. They caused big headaches because the IT team didn't have a solid process for bringing those new groups into IT consolidation and planning. "That was a miss," Mott says. "We didn't understand how disruptive that would be to our process."

Didn't Plan For Success: IT used the sales guidance HP gave to Wall Street, that its best estimate was 4% to 6% annual revenue growth. That forecast mattered, because IT's plan was that all new purchases of servers and storage during the transformation would get delivered to one of the new data centers.

The team didn't want to buy and install systems in, say, a small European data center that was just going to be closed down. That plan didn't work, because several business units--foremost the PC group, which IDC said gained 4 percentage points of share in the first quarter of 2007, compared with the prior year--shot past their goals, forcing IT to provide stopgap capacity in data centers it wanted to retire.

One It Didn't Make: Move the deadline. Given these problems and particularly the unexpected growth, Mott thinks he could have sold an extension of the three-year program, but he knew doing that would seriously damage the technology group's credibility. So he pushed the team, extracting an admittedly serious toll in terms of workload, and they hit the goal.

Besides the credibility effect, it's also against Mott's fundamental philosophy to extend a deadline. "Time doesn't make it better," he says. While the planning team didn't have perfect visibility into the problems that might come up during a three-year program, making it longer wouldn't make things clear, and then there's no telling what else might come up if he gave the team another year.

Like having to integrate EDS, which HP acquired in August? "EDS would be a very good example," Mott says. "Four years would've turned into five."

Return to the story:
HP Goes All In With An IT Transformation Continue to the sidebar:
The Metrics That Matter Most