Peter Weill, the director of the Center for Information Systems Research at MIT’s Sloan School of Management, says IT infrastructure is the foundation on which all e-business depends. Weill is a leading scholar and proponent of the notion that IT infrastructure is linked closely to business strategy-or should be. Weill recently spoke to CIOs about his ideas at CIO Insight’s fall conference in Chicago on IT-business alignment. What follows is an edited transcript of his remarks at the conference.
As I was rushing out the door of my office at MIT’s Sloan School the other day, a finance professor, a colleague of mine, asked me, “Peter, Peter, where are you going?” I told him I was running off to give a talk on IT infrastructure. I’ve worked with this guy for a while, and he told me that after working with me, he finally knew what infrastructure was. I was really pleased, so I asked him to define it. He laughed, and said, “Infrastructure is the part of the IT portfolio that you guys can’t cost-justify.”
Well, my first reaction was frustration. But then I realized we’ve made great strides. Now the finance people are at least thinkingof the notion of an IT portfolio! Seriously, I think a lot of people think about infrastructure that way, and in a sense they’re right.
What we know about infrastructure is that it can be both an enabler and a barrier to business initiatives. So knowing what the business initiatives are likely to be ahead of time will make the job of producing and providing a good infrastructure a great deal easier.
First, let me say that completely flexible infrastructures just aren’t possible-not even with outsourcing, not even with Web servers, not even with ERPs, and actually, I’m not sure we should use flexible and ERP in the same sentence. The idea of a completely flexible infrastructure hasn’t yet arrived, and so we need to know what the likely business initiatives are going to be in the future to help plan the infrastructure today.
Another thing you need to decide about infrastructure is what to share. I can tell you from our research at MIT that infrastructure represents more than half the total IT portfolio in terms of spending. And one of the big issues CIOs have to decide is where to place that infrastructure. Do you place it firmwide? Do you place it countrywide? Do you place it at the business unit level? If you’re an organization like Johnson & Johnson, you have infrastructure at many different levels. If you’re an organization like State Street Corp., we’ve just completed a case study of State Street, you’ve moved from a very independent set of business units to a shared services infrastructure, at least in part. Why? CEO David Spina has announced and articulated a strategy of one State Street, so a new infrastructure is required to provide a single point of contact and also some economies of scale across the business units.
A third point I want to make is that I think there is an age-old tradeoff of infrastructures of all sorts. Firms that invest more in infrastructure than their competitors, according to our work, have the following characteristics: faster time to market for new products and services; more sales from those new products and services; and higher revenue growth. That’s all good so far, but they also have lower ROIs in the first two years after the infrastructure investments. These are all statistically significant results. What that tells us that there’s sort of a brass balance you have to make between investing in infrastructure for today and for the future.
When I tell this to boards and senior executives of organizations, they just nod and tell me it doesn’t matter whether the infrastructure is IT or buildings or people or training. Indeed, these infrastructure tradeoffs are what make management interesting and tough. So we need to recognize that choices are necessary.
Business and technology strategy is about choices, and one of the things that we have found in our research is that different sorts of strategy choices require different sorts of infrastructures. It helps a lot if you know what the strategic initiatives are going to be.
Also, think of infrastructure as a set of services. And of course, infrastructure is also that wonderful complex combination of people and technology and policies and plans all mixed in together, and that’s really the broader rather than a very technical definition of infrastructure.
We think of infrastructure as the base foundation of IT capability, something that’s shared across the organization. These services are relatively stable over time; they don’t change much. A PC LAN service or a shared customer database service or a security service is going to be needed tomorrow and next year and the year after. Even though the components might change, the laptop might change, the service itself will stay relatively stable. Thinking infrastructure as services allows the business folks who pay for this infrastructure a much easier job of valuing that service. It’s much easier to value a reliable PC LAN service than it is a box or a server. And so this notion of service seems to be much more intuitive to organizational managers.
The second thing that I think is so attractive about this notion of infrastructure as shared services is that you can then apply an SLA to it. You can have a service-level agreement that says that’s what we’re expecting to have for that service, and that’s how much we’re going to pay for it. Also, the thing about services is that you can then, if you want, market test that service. You can look for other providers, you can compare and contrast, and that empowers managers in a way that I think is very important. In addition, you can change the technical components almost opaquely to the service over time, so as new technologies come through, the service doesn’t change from the consumer but the technology pieces can be updated as the technical folks see fit.
So this notion of infrastructure as a set of services is very powerful, and I think the biggest growth we’ve seen in infrastructure has been in this shared and standard applications piece where there’s been a real move in organizations. We call it gravity. Dropping infrastructure out of the business units and putting it into shared infrastructures. And so these applications that are shared now are what we think of as shared and standard applications that don’t change very frequently. They’re things like accounting, budgeting, HR. In some organizations the ERP or part of it would be shared in standard, in others it wouldn’t be. It would be different in different business units.
Now this notion of infrastructure allows us to be much more flexible in being able to respond to rapid shifts in business strategy as well. One of the studies we did recently was very scary. What do you think the average life of a business strategy is in a large corporation, let’s say a Fortune 500 company in this country? How long before you make a significant change in that strategy? Five years? Eighteen months? Nope. The average we found was 11.7 months before a significant change in strategy occurred that required new applications to be created. So you need an infrastructure that’s got a stable base that’s going to react, and you also need to have, if you can, forewarning of what those changes are going to look like.
Now, this can get complex. If you’re an organization as large and as complex as Johnson & Johnson, you’ll have 200 business units or so and other levels of infrastructure. They have infrastructures by sector, by country, and they think about what capabilities they have at each level. One of the things Johnson & Johnson has done recently is move to a much more centralized infrastructure than they ever had before, and part of that was a response to their large customers like Wal-Mart and others who wanted to have a single point of contact and not 16 invoices or 16 salespeople from different business units. And so that response has resulted in more of a shared infrastructure, which has very much not been the tradition at J&J and has taken some time to implement, but I think it’s creating quite a great value now.
As you think about infrastructure in your own organization, think about where the capabilities are, and ask yourself how your company decided where those capabilities are. Why? Where you place infrastructure is a strategic question, not a technical question. If you find yourself answering the question for technical reasons, I suggest you take a look at it again. It’s a business decision where you place infrastructure capability because that’s where you want to use the results of it.
Let’s look at an example of an organization that has made amazing strides in infrastructure. At Delta Air Lines, they’ve moved from a very fragmented infrastructure to the Delta Nervous System, and it’s been a remarkable turnaround. I’d have to say that there’s still work to do, and it’s been a terribly difficult year or so for the airline industry, which has put a lot of pressure on their infrastructures. But I think because of that and because of their turnaround, we can learn a great deal from the way in which they’ve approached it.
First, some background. Delta is the third-largest airline, a very large organization, more than 60,000 people. Leo Mullin, the CEO, joined Delta Air Lines in 1997, and when he joined-now this is a tough organization-Delta was last in three of the key metrics that are published regularly and very important in the airline industry. Delta was last in on-time departures, customer complaints and baggage problems.
One of the reasons, not the only one, for these problems was Delta’s infrastructure. At the time they had 17 independent platforms, and Charlie Feld, who was the CIO who led this change in infrastructure, commented to Leo at the time that Delta just couldn’t get from here to there on Y2K. So one of the big motivations for this change was Y2K. There were 17 platforms, one for everything-one for flight operations, one for baggage, one for crew scheduling-and there are many other organizations that had a similar structure.
So Delta rebuilt its infrastructure around core business processes, and that’s a very key lesson. They identified the four core business processes. The main motivation, and I think you always need a motivation or a metric for infrastructure, was to enhance customer experience. So now they have a multichannel infrastructure that serves multiple segments. They’ve had some very successful investments, such as in priceline.com[deleted per Marcia], and very successful use of delta.com, and you might have seen recently their changes in pricing to encourage more people to use delta.com, which is putting pressure on the travel agents in the channel.
So what they ended up with was this thing called the Delta Nervous System. It’s been a billion-dollar investment, and at the time, and I wouldn’t hold them to it now, but at the time they committed at the board level and senior management level to renew that infrastructure every four years. So reinvest a quarter every year. Now, of course, in this climate that’s not going to be possible I expect, but that was part of the thinking to make this infrastructure grow and be important over time.
Infrastructure tends to be lumpy, it comes in chunks, and one of the great lies of the 20th century was told to me by the CEO of a bank that you all know and love. He said, Peter, you know, I get told by the IT people every year, when they come up with a huge $50 million infrastructure proposal, that we won’t need any more infrastructure for a while after this one. And that’s one of the issues that you have to manage over time.
So three years later, Delta has turned around the business, and it made billion-dollar profits two years in a row and also led on the three key customer metrics. That’s an amazing turnaround, and it’s not just infrastructure. Of course, there are many things that contributed to that, including Mullin’s leadership.
In 2002, Delta led with some innovative customer service features-pages, for example, to tell you if the gates have changed. You can print out your boarding passes at home if you’re a certain level of Delta member. Things like this.
This deceptively simple picture, Delta tells us, took 60 iterations over four months to get right, and I’m sure it’s changed since then.
In fact, before the Delta Nervous System, if you went to four different people in Delta and asked for a gate for your flight, you may well have gotten four different answers because they were looking at four different systems. This was very frustrating to the people who worked at Delta, as well, because they knew that the irate customer standing in front of them was irate partly because Delta employees couldn’t do their jobs.
So they rebuilt their infrastructure, if you will, to improve customer experience, to build an operational pipeline—and in an airline it’s all got to be connected and synchronized, so you can see all the pieces. Delta also wanted to create a better revenue management organization and a revenue management architecture that sets prices, sets fares. It’s an incredibly complex world setting fares in a large airline.
Now in the middle of this new infrastructure are nine databases, and that’s really the heart of the Delta Nervous System. Many of them existed before Delta made this change to a strategic infrastructure; some of them they’ve updated. And then they’ve built a publish-and-subscribe environment around that. They used TIBCO and other tools, and created an environment so that if, for example, you’re a baggage handling manager, you can subscribe to events that occur, and then every time that event occurs, like a gate change, that will be updated in your systems. And it’s been very successful for Delta.
Indeed, the whole notion of Delta’s Nervous System is really based on having a core infrastructure that ran Delta Air Lines from a real time perspective. It was all about data, really, getting the information about Delta Air Lines, where are customers’ flights, where are Delta’s crews, where’s the schedule in terms of changes. How do you take that as well as all the middleware and all the basic computers, how do you connect that into something that reflects the true nature of the airline so that instantly, as soon as a plane changes its status or a crew changes, everyone who needs to know is aware of that?
And so that was really kind of the base thought behind Delta’s Nervous System, and that was the infrastructure that was put into all of Delta’s airports and their rest centers.
The Digital Nervous System allows Delta to make a change around a flight, let’s say a cancelled flight, and with one or two entries, that information would be pushed into all of the operating and customer groups without an individual or 25 individuals having to actually access that information and send it.
The information would come to the reservation call centers, it would go to the airports, it would go to the Crown Room Clubs. It would actually go to PDAs and cell phones and beepers. It would go to customers’ laptops, giving them information around the fact that Flight 222 from Washington to Atlanta has been cancelled, and that Delta has rebooked someone on Flight 223 that leaves in two hours.
Saying you have a nervous system that you’re going to connect all your body parts to is a wonderful idea and the concept is exciting, but there’s a tremendous responsibility from a reliability, stability and backup and recovery standpoint. And the good news is everything will have access to everything in real time. The bad news is if it doesn’t work, if it can’t scale, if it gives you the wrong answer, then literally you shut down the airline.
Now, even though Delta was spending hundreds of millions of dollars on the new infrastructure, it was spending that kind of money anyway. That’s the fallacy in what most people think. When it’s being spent in departments and in divisions, the money is being spent. It’s just not being seen.
The former Delta CIO Bob DeRodes’ main complaint with this new system was that the list of applications was long and there was so much value in them, that it was both a challenge that he had to deal with and a burden that he had to carry. But, my goodness, I mean, think about how things were done in the past. According to DeRodes, three years ago everybody at Delta was sitting back on their thumbs, and Delta was way behind, and the performance of the company reflected it.
I think Bob described it the best by saying his job with regard to infrastructure was not to win the war but to build a civilization. Bob has been very articulate, understands the airline business, but, more importantly, understands things like customer service, understands things like high transaction processing, and he will take and with the team that’s in place actualize this vision of the digital nervous system. And I think this is where the rich are going to get richer.