MasterCard Is Charging Ahead

With 65 percent of corporate revenues derived from IT initiatives, MasterCard International Inc.s systems are highly strategic, especially in its ongoing quest to best archrival Visa International Service Association. But getting to this point wasnt easy. It entailed a massive eight-year rearchitecting of the global payments companys data center and ne twork. The recently completed undertaking, called System Enhancement Strategy, also included a total rewriting of MasterCards applications.

Leading the SES effort for MasterCard were Jerry McElhatton, president of global technology, and Robert Reeg, senior Vice president of systems development. Both men met recently with eWEEK Executive Editor Stan Gibson at MasterCards technology headquarters in OFallon, Mo., outside St. Louis, to explain how they made a point of avoiding outsourcing and relied instead on disciplined management.

When MasterCard decided to overhaul its systems with the System Enhancement Strategy initiative eight years ago, did you feel you were falling behind rivals like Visa and American Express [Co.]?

MasterCards IT assets OFallon, Mo., data center 550,000 square feet; cost: $136.9 million IBM zSeries mainframes Two in OFallon; one at backup center elsewhere in the Midwest; one in Europe Sun servers About 300 Distributed systems 684 IBM NetFinity systems, called MasterCard Interface Processors, allowing members to securely connect to the MasterCard network Data warehouse 80 terabytes Storage EMC Corp. storage area network environment; 271 terabytes Application development environment Primarily Java and C++ Database platform Primarily Oracle Resource Library: Desktop environment Microsoft Windows and Office applications, IBM Lotus Notes Networking AT&T virtual private network IT staff at OFallon data center 2,300 Project in progress Converting by 2005 Europays systems, based in Brussels, Belgium, to MasterCards common architecture—about 10 percent of MasterCards current transaction volume; Europay acquired last year

McElhatton: I dont know if it was so much that we were falling behind. We wanted to make sure we could go into the future and provide the functionality and capabilities that would support both MasterCard and our members well. I think our systems were very competitive, but they were very mature systems. And we didnt have some of the capabilities to be as responsive as we wanted to be or needed to be. It was like a pickup with 300,000 miles on it—it may still run, but at some point there are better, newer and faster things. We want to make things faster, better and cheaper. Today, MasterCard generates about 65 percent of its revenue through technology services. When you generate 65 percent of your revenue from your systems, youve got to make sure youve got the right systems in place.

Can you quantify a payback to SES?

McElhatton: We get asked that often. Its hard to put in dollars. We wanted to make sure that our capabilities would allow us to continue to do what we do well but, going forward, would also give us improved time to market, new capabilities, new features and new functionality.

We also wanted to make sure that we could give our members the utmost in capability to develop potential revenue, taking maximum advantage of our system. We think that weve got some competitive feature functionality, and we think weve got the ability to deliver things very quickly. That gives us a real competitive advantage.

Are you working to raise the percentage of IT-generated revenue, say, to 75 percent?

McElhatton: Yes. In addition to our core transaction processing services, we have user pay services. Risk products are a good example. A risk product is designed to reduce fraudulent transactions at the sites of our member merchants.

Reeg: You could have a neural network that looks at a transaction and scores it according to risk factors and then gives the score back to the members.

McElhatton: Weve got products called Safe and Match, where we know if there are merchants that have been fraudulent in the past. Another area were looking at is "bust-out" schemes, in which people apply for, say, 20 credit cards, run the bills up and leave the country. We also have the Issuers Clearing House System so we can make sure were not going to send the card to a penitentiary or a vacant lot. We either sell our fraud or risk products to members or run the applications on behalf of the members.

How do you use data warehousing?

McElhatton: Our data warehouse is about 80 terabytes. We store transactions online for three years, and we have near-line storage after that. We have a suite of products for our members thats significant. The members may subscribe to a service so they could use the data to see how they are doing against a peer set. They may use the information to offer a special promotion. Those are user-pay services, and thats where a chunk of the 65 percent of revenues comes from.

What did SES cost, and how did you manage the project?

McElhatton: In this project, a $160 million project, we were on time or ahead of schedule and within budget. We do a very good job of managing major projects. Thats a big part of our business. We have tremendous people who have tremendous confidence in their capabilities.

At the same time we were executing SES, we handled Y2K and moved to our new building here. Were also building a contingency site elsewhere in the Midwest that has just been finished.

Were also working on convergence with Europay and its EuroCard Systems.

We have the capability to do this stuff on time. Rob and his people use good discipline. We get good estimates from people, and we manage scope creep.

Reeg: SES was an opportunity to get on the same platform with the same tool set. We were very careful about setting the focus. Even though not everyone would agree that was the right direction, we made a decision that was going to be the way that everybody marched.

We also broke it down into bite-size chunks, so we would not have analysis paralysis. We made sure there were twice-a-year deliverables from SES that were measured and budgeted for. Keeping that focus and keeping everybody aligned on the delivery was one of the keys of success.

McElhatton: You cannot do a $160 million project by throwing $160 million at it. Youve got to break it down into small, manageable pieces. The industry average for on-time projects is about 30 percent. Were over 90 percent.

Are you using project management software?

Reeg: We have an in-house methodology that we call "master project," and we use a variety of tools, from Microsoft [Corp.] Project to Niku Corp.s Niku Workbench, to some in-house tools. That gives our project managers detailed information on the status of their project.

So everyone is working off a common platform?

Reeg: Yes. When we made a definite decision that we would work off a common platform, there are pros and cons with that. MasterCard prior to that had operated on the basis of whatever [platform] fits a particular project. That was good from a project level but not so good from a cost and efficiency level. So we have a common platform across the organization. Rolling all the way up to Jerry, theres one common view of every project.

Do you use consultants at all?

McElhatton: Not a lot, but we have a group, DiamondCluster [International Inc.], in right now looking over our shoulder. Its interesting. They found a lot of good things and some areas of improvement.

Do you use offshore outsourcers or have offshore operations?

Reeg: We have an operation in India. In 1999 or so, there was a crunch on getting programmers in the U.S. You couldnt find enough skilled people. So we formed a partnership with a company in India called Mascon-MasterCard Global Technology Services [Ltd.] to help bring skill sets to the table. Other companies may be motivated by cost, but we needed more resources. And India has great technical resources that can augment the very superior technical staff we have here in St. Louis. Thats turned out to be an incredible partnership that brought a lot of talent to the table and saved MasterCard a lot of money.

Do you use any H-1B workers here in the United States?

Reeg: Yes, we do. But we dont outsource. We integrate our India people and our on-site people as a team. At any given point in time, 40 percent of our joint venture in India will be right here in St. Louis.

Why do you want to stay away from outsourcing?

McElhatton: Formerly, some of our debit processing was outsourced. We brought that in-house, and our costs went down about 350 percent, and our quality went up significantly, and our time to market improved. We feel at MasterCard that our products and services are so key to us that we have to control them. And we run a very cost-effective shop.

One of the things you have to remember about this business—if you ever go to use your MasterCard, you dont want it ever not to work. We dont want it ever not to work.

If you try it and it doesnt work, youre going to stick it to the back of your wallet, and youre going to pull out a competitive card. We dont want that to happen.

We have tremendous redundancy built into our systems. We dont want to be one of many. We want to control our own destiny. And technology is so important to us because its such a large generator of our revenue.

Was the new disaster recovery strategy and the new backup data center spurred by the events of Sept. 11, 2001?

McElhatton: There were several reasons we did it. On 9/11, we had two applications, settlement and debit, running at our backup site in Lake Success, N.Y., due to hardware and software upgrades at our main site. I was in New York and had to drive back to St. Louis. It was 18 hours of driving time.

As a result of that, we looked at our building in Lake Success. It was not built to be a data center. You could drive right up to it. So we started to look at five different locations and chose another Midwest location, which I would rather not name. It was a computer center before, and it was in great shape.

Are you satisfied with your architecture the way you have it now, for at least the next five years?

McElhatton: We just came off of a cycle that lasted about 20 years. Now we have a lot more flexibility in terms of databases, code, scalability and portability. We have two major software releases per year. This gives us the capability of doing things between those releases. It gives us a tremendous amount of flexibility.

I think this system will last us a long time. We can make enhancements to it as we go along.