What Stays and What Goes

The vetting process was only the beginning. Doucette also had to create an overarching strategy for outsourcing to India. Taking the time to make such a strategy clear is something novice CIOs might skip, but that would be a mistake. "There's much more to going offshore than sending out an RFP, selecting a vendor and doing it," says Marty McCaffrey, executive director of Software Outsourcing Research. "You have to go to extraordinary lengths to establish goals and objectives first."

Doucette created a sourcing plan stating what goes offshore, what goes to US outsourcers and what stays at UTC. His criteria for what goes where are pretty simple. "First, you have to ask yourself if the work is strategic. If the answer is yes, you should keep it internal," Doucette says. "Then, if it's not strategic, you have to ask yourself if it's going to the lowest-cost source. If you're not the lowest-cost provider of that service, you need to contract it out."

The result of that analysis: Doucette and his divisional CIOs send all help desk, network, desktop, midrange, mainframe and Web-hosting work to Computer Sciences Corporation (CSC), and have a goal of sending 80 per cent of all application development and support to India. That means UTC and its subsidiaries keep in-house IT leadership, project management, business analysis, and (eventually) 20 per cent of application development and support. Keeping a certain number of skilled developers in-house ensures that UTC and its divisions don't get overcharged by an outsourcer. "When I was at Otis," Doucette says, "we put out an RFQ for an e-commerce project, and we got bids ranging from $US60,000 to $US3 million. If you don't have someone in your operation who knows how much that should cost, then how can you outsource it?"

It's a mistake many CIOs make. "One way to circumvent that is to have multiple vendors continually bidding against each other," Meta Group's Davison says. "Another way is to keep people in-house who are really smart about those things."

In setting standards for what should go to India and what should stay, Doucette also took into account that UTC's six subsidiaries have very different structures and needs. Companies such as Otis, with most of their IT departments centrally located at headquarters, were good candidates to go offshore. Carrier, however, was not. It has 49 manufacturing plants scattered across the US with a handful of IT people located at each one. "When creating an offshore model, you really need to look at your own business model and where your IT staff is, and customise it based on that," explains Doucette.

Another issue Doucette had to consider was the companies' security needs. While security is clearly a priority for every UTC company (it uses secure private networks, stores most code stateside and uses software that limits offshore network access), UTC simply cannot put any military-related work offshore. Pratt & Whitney, Hamilton Sundstrand and Sikorsky Aircraft all do government work, which accounts for 16 per cent of UTC's combined revenue. "We have to split up military and commercial work and be careful about what we give them access to," Doucette says. "It's a struggle because we're trying to work together as much as we can to save money. For example, if Pratt & Whitney and Carrier are in the same building, they share a network. But that means I can't put that network support in India even though I'd like to."