GM's IT Overhaul: 5 Critical Questions

CIO Randy Mott's plan to overhaul General Motors' IT operations drew a wave of responses from readers. Some cheered the insourcing and new U.S. development centers that are planned. Others fretted about the IT bureaucracy that could be introduced. Some jeered the strategy and Mott himself.

Mott's IT "transformation" plan, as it was at Hewlett-Packard previously, is high risk, packing major changes on multiple fronts into a three-year window. And Mott is doing it while GM tries to bounce back from a bankruptcy filing and government bailout, so IT is far from the only place at the company where things are changing fast. "GM is in the process of making a lot of decisions," he says.

Some readers called Mott's strategy "really nothing new," since ideas such as consolidating applications and data centers and measuring IT results aren't revolutionary. That's one I can't agree with. You can debate the merits of the strategy, and we'll see if it delivers. But I don't think anyone at GM will feel like nothing changed in the company's IT.

[ Want to hear Randy Mott discuss his strategy live? Read Register for the InformationWeek 500 conference. ]

I've interviewed Mott multiple times during his stints as CIO at Dell, HP and now at GM, and he has only grown more adamant that CIOs must identify what needs to change in IT--and then change all of it at once. Transform the whole organization, he argues, or the pieces that remain unchanged will hold you back. (My colleague Rob Preston will interview Mott live at the InformationWeek 500 conference in September. Register here to come and ask your own questions.)

So even as GM looks to hire hundreds, even thousands of IT pros as Mott dispenses with outsourcers and brings application development and other operations in-house, GM will consolidate its 23 data centers into two, more-automated ones and reduce the 4,000-some apps GM uses by 40% or more, so that those new staffers can focus on new IT projects instead of just supporting legacy infrastructure.

That all-in approach ramps up the risk, by pushing so much operational and cultural change on the IT organization and the company. But Mott thinks it's also why top technologists will consider working for this recently bankrupt business. "A lot of IT people are very driven by this, they're driven by a sense of accomplishment," he says.

The journey will be hard. Listening to readers' comments, both online and via email, here are five fundamental questions on your minds, with additional input from Mott based on our interview in June.

Can GM keep up IT performance while cutting apps and data centers?

Mott wants to consolidate down to two main data centers, equipped with the latest, most automated hardware and software. He wants to reduce the number of apps--for tasks from monitoring inventory and checking warranties to tracking consumer trends--by making more of them the company's global standard. About 30% of GM's apps are global today, and Mott would like it closer to 80%.

One reader, "HP Survivor," says HP's network and apps suffered from the consolidation and application cuts Mott instituted there: "The bloated and inefficient bureaucracy created by that 'transformation' hampered development tasks and crippled R&D teams ... This disaster was wrought by the brutal insistence that IT spend should be at or less than 1% of total revenue."

Global CIOs: A Site Just For You

Visit InformationWeek's Global CIO -- our online community and information resource for CIOs operating in the global economy.

Two points here. One major difference between the IT overhaul at GM and the one at HP is that GM isn't pushing for big IT cost cuts. Mott's goal at HP, at the direction of former CEO Mark Hurd, was to cut IT spending as a percentage of revenue from 4% to 2% by 2008.

At GM: "Will we lower the budget? Yes, but it's probably a little as opposed to a lot," Mott says. "What we do want to change is where we spend the money. We don't necessarily feel we're terribly overspent like in the past example you brought up, but we're definitely under spending in innovation and overspent in running the business. We really want to save dollars in the run side and invest those dollars in the innovate and build side of supporting the business."

It's entirely true that infrastructure performance--data centers, networks, applications--are the IT table stakes. This new IT org can't be an innovation engine if engineers don't have the essential apps and bandwidth they need to craft cool cars, or designers and marketers don't have the data they need to assess trends.

Can GM get its data in order, to better understand customers?

GM CEO Dan Akerson has said that GM needs a much stronger data warehouse capability, so that employees can use data to better understand customers, market trends, and business performance. "We're going to be very, very data-driven rather than episodically driven," Akerson told Fortune's Geoff Colvin.

At HP, Mott in 2008 said the company had gone from as many as 700 data marts in 2005 to 50, with a plan to get to one central data warehouse. HP IT was the marquee customer for its own data warehouse product, called Neoview, which it was pitting against the likes of Teradata and Oracle. But HP has since dropped Neoview. Wrote one commentator: "What is HP doing now that the infamous Neoview is no longer in development?"

Mott is intentionally vague about what GM will implement to consolidate its 200 data marts, because the IT team is sorting through the technology options. Mott plans to decide on those choices by year's end. Getting to one architecture might not mean, literally, getting to one data warehouse or one technology, but the goal is to make all data centralized and therefore easier for employees to access and share, he says.

This strategy could create some tension with GM employees. Mott loathes "shadow IT," where business units set up systems or data stores without IT's help. But he recognizes that his colleagues all had some good reasons for going after the data they need. Mott says it's not accurate to say that the business doesn't know the really critical pieces of information they need to do their jobs, because they've worked hard to get that information. Says Mott, "The question is, would they know a lot more if they didn't have to work so hard to get it. The answer is yes."

Will Mott's IT planning effort be seen as efficient or bureaucratic?

Mott requires a cost-benefit analysis for every IT project. If it's a small project, it's a small CBA, he insists. Mott uses a rigorous planning cycle that forces business unit leaders to prioritize projects, so IT is putting the most important ones first.

This process drew many doubters among readers. Says one: "The juice, generally, isn't worth the squeeze. Yes, you can probably hammer down a fairly comprehensive ROI for projects if you have a dozen analysts working on the numbers, DCFs, NPVs, etc, but, except for very large projects, the cost of the analysis and the bureaucracy consume the gains. All of this analysis on ROI also makes it really complex to do anything, as building the business case is such a hassle."

Another commentator who says he was a former GM senior IT staffer says GM had that problem in the past and needs to get more nimble: "IT itself has bureaucratic processes. ... Each project has to provide so many metrics and dashboards that are of no value."

I've pushed Mott on this topic several times, asking if the overhead of his planning and reporting effort is worth the time, and whether it risks slowing the speed of execution. After all, one big reason for insourcing is so that GM can move faster, and not start each new IT project with a contract negotiation with the outsourcer.

Mott thinks the CBA process is worth it for many reasons. One is that it forces the business units to tell IT which projects are more valuable than others. That is, it recognizes that IT is a scarce resource, just like financial capital, so not every project will get done. It lays out, with the finance team, what the benefits are.

Mott also wants to know what the IT organization accomplished. It's how the company knows if IT is delivering, and it's how IT pros will be motivated to do the unglamorous work ahead, such as consolidating data centers and apps. "You have to drive a high sense of accomplishment," Mott says. "If you're going to drive a transformation, you have to measure it, and the whole organization has to know what it accomplished."

Mott knows this CBA process takes extra effort, especially while people first learn the process, and people will support it only if "by doing it people see they actually get more."

Where will GM put its new software development centers?

GM plans to have four U.S. development centers, including one already in place in Warren, Mich., near Detroit. Mott says the other three locations are to be determined and will be based largely on the ability to recruit tech talent, including from universities.

We couldn't help but speculate that one would be in Silicon Valley. It's the country's most expensive market for IT talent, but companies such as GE and rival Ford have decided they need to be there to stay on software's leading edge. One reader wrote me to question that wisdom: "Sure, it's not too far from either UC Berkeley or Stanford, but after that, what? The East Coast has a denser fabric of university talent." He offered Boston as a promising locale--though it's also expensive and isn't nearly the tech hub Silicon Valley is.

The development center locations will both reflect and influence the culture GM's building. Is it Silicon Valley and Boston, or is it somewhere like Pittsburgh, Charlotte, Dallas, or Denver? Warns the reader: Silicon Valley "is famous for its mercenary culture, and it would seem by insourcing, Mott is trying to (re)create a corporate culture of loyalty. That doesn't fly in Silicon Valley."

Can GM hire well and build a new culture?

This to me is the single biggest challenge. For GM to shift from 90% outsourced IT to 90% in-house, it needs to hire hundreds or even thousands of people, assimilate them into GM's post-bankruptcy culture, and rewrite GM's IT process along the way.

The personnel shift is the thing that excites a lot of commentators on our site, who tend to be anti-outsourcing: "IT is not a luxury but a necessity for any company to run effectively. Handing over the company's data to an outsourcer is plain unadulterated stupidity. GM has made some bad decisions. They are finally doing something smart."

A commentator on our article "General Motors' Big Bet On U.S. Tech Talent" sees great promise for GM in hiring during a weak job market: "GM and other employers can now shop for top IT talent from the best schools, plus semi-experienced IT pros whose incomes have been reduced due to [outsourcing and the bad economy]."

But changing a culture is fraught with risk. Mott clashed with HP's entrenched culture at times. And the scale of the cultural and HR challenge is daunting. For a sense of it, read Allstate CIO Jim Ditmore's InformationWeek column on "Why Outsourcing Fails". Ditmore was on the team that brought IT in-house at Bank One, under CEO Jamie Dimon and CIO Austin Adams, and he says it took hiring 2,200 people and revamping processes and infrastructure over 18 months.

Writes Ditmore: "This effort aided the bank's turnaround and allowed for the merger with JP Morgan a few years later. As for having in-house staff do critical work, Dimon said it best: 'Who do you want doing your key work? Patriots or mercenaries?'"

But first you have to find and inspire your patriots. Employees are mercenaries, until you unite them behind a cause. Doing that will be the biggest test of all for Mott and his IT leadership team.