S+B: What dimensions should be considered as executives think about how to approach innovation execution?
GOVINDARAJAN: Let’s take a concrete example. In 2000, Infosys [Technologies Ltd.], which had been writing software, went into the consulting business. Infosys Consulting could not be launched inside its core business. The performance engine was designed for software development, whereas consulting is about business transformation; programming and consulting require fundamentally different capabilities. And the customer for custom software is the head of IT, whereas the customer for consulting is the CEO. So a dedicated team was created, and the team designed a different culture and compensation package for the new business.
At the same time, the dedicated team drew on certain assets of the performance engine. One very important one was customer relationships, because, after all, the consultants wanted to work with the same Fortune 500 companies as the programmers. A second asset was the existing global delivery model, which enabled Infosys to dramatically lower costs and still offer high-quality service.
S+B: How do you ensure the performance engine’s participation in executing the innovation?
GOVINDARAJAN: You have to recognize that there are fundamental tensions between performance engines and innovation teams, and that any time you force them to work together, conflicts will arise. So you have to make the link, and then expect the conflict and manage it. The Infosys case study revealed three ways to do this.
The first is to have the right leader for the dedicated team. At Infosys, it was an outsider named Steve Pratt. That is important — he wasn’t part of the performance engine. He also happened to be an extremely humble person, who was willing to talk through the potential conflicts with the performance engine and reconcile them ahead of time. And he didn’t overreach. Instead of asking the performance engine to give him a lot of customer contacts, he asked for one customer contact and proved he could sell consulting to that customer without damaging the programming business.
Second, the dedicated team, like the performance engine, should report at a very high level. When Infosys Consulting was started, it was a tiny business. Yet it reported directly to the then CEO, Nandan Nilekani, who played a critical role in anticipating and managing the conflicts between the performance engine and the innovation team. Nilekani also created another board of directors for Infosys Consulting and made sure a few members served on both boards.
Third, it is important to create incentives, because ultimately people respond to rewards. One of the things Infosys did was to double count the revenues from the new consulting business. So if the software business provided a lead that generated sales for Infosys Consulting, it also got credit for the new revenue. This gave the performance engine an incentive to work with the innovation team.
S+B: Can you evaluate an innovation team in the same way you evaluate the performance engine?
GOVINDARAJAN: No, remember that innovation execution and day-to-day execution are fundamentally different. The performance engine is rightly held accountable for tight, financial-oriented metrics. But you cannot do that with the innovation team because it is conducting an experiment. That does not mean there is no accountability. We can hold people accountable for disciplined experiments.
Innovation projects start with nothing but a bunch of assumptions about the future. I call them weak signals. When you are faced with weak signals, you have two options. You can wait for the signals to become crystal clear before you commit resources, but by the time the signals are crystal clear, the game is usually over. The second approach is to amplify the weak signals yourself by testing the assumptions and converting them into knowledge. When you are converting assumptions into knowledge, the golden rule is spend a little, learn a lot. You evaluate the innovation team by how well and how quickly it amplifies weak signals.