They said he couldn’t do it. After all, ceding control of key processes of one of the world’s biggest multinational companies to India — a country whose economy was shaky and whose people were mostly impoverished — seemed a ludicrous idea at the time. But in the mid-1980s, Raman Roy, who back then worked for the accounting division of American Express in India, quietly showed the travel and financial services giant that using the talent pool and resources of the world’s largest democracy would be a strategically wise move that could save the company lots of money without compromising quality. After leaving American Express, Roy did the same thing for General Electric before launching Bangalore, India-based Spectramind which was acquired in 2002 by Wipro, an Indian software giant. In a recent discussion with Knowledge at Wharton and Ravi Aron, professor of operations and information management at Wharton, Roy shared the story of how he built an upstart outsourcing firm into a leading provider of services for some of the world’s largest companies. Below is the first part of the conversation; the second will appear in the next edition of Knowledge at Wharton.


Aron: You began your career as a consultant but were tapped to join American Express in the mid 1980s. What did you learn at that time about Indian operational and executional capabilities?


Roy: When American Express launched its Indian rupee card, some of the things we had to do at the company’s Indian division were similar to what was being done in the more advanced markets. Our operations began to resemble the operations shops of the other company units in terms of servicing, number of transactions, complexity of transactions, and so on. As we did that, an interesting fact emerged: the cost per transaction from India was significantly lower than those of the more advanced markets. And the quality of the output, which was a function of the quality of our staff, was significantly superior.


People in the company began to talk, saying, “These guys in India obviously don’t know how to make their reports; how could these numbers be true?” But when senior managers started working with us and reviewing the process, they realized that we did know how to make the reports – and the actual cost was low.


Aron: How much lower were the costs in India?


Roy: They were lower almost by 40% to 50%. But lower costs don’t tell the full story because the gains were more than just in cost savings. The real gains were greater because of our execution efficiency, the number of first-call resolutions of customer problems, and the ease with which we resolved the issues (which was purely a function of our more qualified and better educated staff).


All this caught the eye of John McDonald,  who the comptroller at American Express. It planted a seed in his mind: He found our Indian operation interesting — especially our cost base and the quality of our manpower. Later John decided to consolidate the company’s multiple operational centers into three locations. He chose Phoenix, Arizona, for the Americas; Brighton, U.K., for all of Europe; and he was on the lookout for a center to serve Japan, Asia/Pacific and Australia. The seed that had been planted many months earlier in his mind came to fruition, and he nominated some of us to work with the group that was evaluating locations in Japan, Asia/Pacific and Australia.


Aron: What did you find?


Roy: We did a lot of research — reviewing government plans, telecommunications infrastructure, what was good and bad about the business environment, and so forth — in each location. Ultimately our recommendation to senior management was that India would be the obvious choice. That’s how the center came into being, and I had the honor of playing a role in leading that effort for American Express.


Knowledge at Wharton: What kind of reservations did people have about setting up operations in India at that time?


Roy: The reservations were primarily due to technological infrastructure issues. For instance, IBM was not in India. Even if you imported IBM machines, local IBM representatives were not there to support the equipment. Say you decided to put an AT&T/Lucent switch on your telephone, AT&T/Lucent was not in India. We realized the barriers that we had to scale because the country had hardly any infrastructure in technology, tech support, or telecommunications. When we went to [Indian telecom firm] VSNL and said that we wanted 100% up-time, they looked at us as though we were from some other world because 90% up-time was considered outstanding. In India, one learned to live with 10% down time, but you couldn’t run an international center like that. Electricity was a problem, the hiring of people, unions, there were various challenges. These were the problems with which we had to grapple.


Aron: Let’s talk about people. We often hear of the superiority of Indian human resources. What was your experience?


Roy: This was purely an accounting center, dealing with the controllership function. So we’d gotten some statistics about the availability of CPAs and chartered accountants and cost accountants. People discovered that you could get a full CPA or a chartered accountant in India for marginally more than what you would pay for an associate in the US.


Aron: Tell us a little bit more about the comparative profiles: who is an associate and what is the profile of an associate?


Roy: An associate would be, let’s say, the person who does the bank reconciliation. In the more advanced markets, such an employee may have just completed high school (or perhaps not even that) and may work for $15,000 to $18,000 or perhaps $20,000 at the high end. At those salary levels, I was able to field professionals who were at least college graduates and who sometimes were CPAs.


Aron: How successful was this initiative?


Roy: It was phenomenally successful for American Express. One aspect was that the shared service center had the strategic intent of reducing costs. We were very successful in meeting that objective. We were able to execute processes that people thought were extremely difficult or very time-consuming. I often thought that it was almost like a pact between John and our management team; he said “cost cutting is all very good, but you’ve got to show what superior work you can do.” And we were able to execute processes smoothly that people thought were extremely difficult, or those that were very time consuming, such as processing open items in a reconciliation. We brought down those costs dramatically, as we brought down the turnaround time in the payment of invoices where there were incremental discounts that American Express could get (for prompt settlement). In fact, I remember making a presentation in which I just took the incremental discounts that we were able to get for the company for prompt payment and compared that to the salary bill of the unit that I used to run for American Express. I told the company, “You’ve got us for free because the incremental discounts that we got alone were more than the salary bill in India.”


Aron: Was there a technology aspect to your success or was it just the better trained human resource delivering the gains?


Roy: Let me give you some anecdotal insights – it is quite an incredible story. As we got the reconciliation of the markets, I had about 12,000 accounts that I would reconcile out of our unit in Delhi. One of the things we discovered was that all our reconciliations ran on the Lotus 123 software. It is very similar to how one would use Excel today, where you match two items and do a ‘line delete.’ But if you took a sip of coffee or someone tapped you on the shoulder, and you deleted the wrong line, the account went out of reconciliation. We had a dire need to bring in some kind of a control and run it off a robust platform.


So the three units in Phoenix, Brighton and India put their heads together and asked the American Express management for a standard platform. The company decided to invest in creating a platform for reconciliation. But by the time the implementation rolled out to India, it would be 15 months and that seemed a long time away. I needed some interim solutions.


I worked with a local company to develop a sort of stand-in reconciliation platform. Incidentally, the head of that company is now the head of technology for Wipro-Spectramind. Sunil Gujral, who had earlier worked for American Express as my head of technology, had left to set up his own company. I worked with him to develop this reconciliation platform. It was meant to be a stepping stone to the ultimate reconciliation platform. About six or eight weeks later, I got an e-mail from the U.S. office saying that the reconciliation package that had been developed did not perform as fast as expected on test data. The effort for developing a new platform would start all over again, and it would now take much longer than the 15 months that we had originally anticipated.


A wild but exciting idea came to me. I asked for the test files of data to be sent to me and ran it on our home-grown reconciliation tool. It took only 25% of the time to reconcile the file. We published those results and soon a bunch of managers and a few technology guys showed up in India to check and see if what we were claiming was true. Within three months, this inexpensive, home-grown tool was nominated as the global reconciliation platform for American Express as an enterprise-wide standard. To the best of my knowledge, it continues to be the platform on which the company handles reconciliations even today.


Aron: How did you make the move to GE?


Roy: GE found out what American Express was doing and thought it was interesting. I became part of the GE team in 1996 after meeting a good number of GE leaders, and so I believed there was buy-in from the leadership. It was a rude shock to see that just because the top leadership bought into it, it didn’t mean the various business units did. The businesses had separate leaders who made independent decisions.


The first project we took on at GE was something they had outsourced to Mexico. And I had to quote a price that was lower than Mexico’s to show that we could do it in India. My CFO came back and said, “By the way, we will probably lose money on every transaction.” But that was the starting point to demonstrate to GE what could be done out of India.


The concept that Jack Welch used when he introduced Six Sigma at GE was to ask each of the businesses to nominate one location as a Five Sigma beta site, where the concepts of Six Sigma could be incubated and fine-tuned before they were rolled out across the company. My unit was nominated as the Five Sigma beta site for GE Capital and we used the learning of Six Sigma, combined with the available talent and technology, to demonstrate what could be done in India. We turned out much better quality on some assignments than was thought could be achieved.


Aron: When you say that you turned out better quality, could you give some examples of how you improved upon the existing benchmarks?


Roy: Errors per hundred and per thousand, turnaround times, accuracy, better understanding of what the customer wanted and delivering that, promise to pay versus actual payment – we judged the quality of our output on all these parameters, and of course the parameters differed from process to process. I found that, in general, we delivered superior quality and bettered existing benchmarks on many of these parameters.


Aron: Did you do any non-batch, real-time work — such as voice-based processes?


Roy: An opportunity came along to take our work to the next level — to find out if the Indian workforce could service online transactions as well as they could handle batch transactions. I explained my intention to experiment and do call centers out of India, and one of the very senior people pulled out his pen and crossed out the investment. He told me that I was smoking something strange, and the company did not see any reason to finance individual fantasies, since, in his opinion, “call centers would never happen out of India.”


I gritted my teeth and told myself, “I’ll show you what can happen out of India.” It was nationalistic pride that drove me to see that a call center came into being. I had no budget to experiment. But four members of my management team and I were fired with a kind of missionary zeal to prove that what we could do in India. So we went to the vendors — the equipment manufacturers — and proposed that they loan us the equipment for three months. If our project worked, the order would be worth 500 seats and it was theirs. If it didn’t work, they would take back their equipment, and we wouldn’t have to pay for it. At that time, it was our ambition to have a 500-seat center. We identified a customer who funded part of the needs for the proof of concept. So, with this one believer that was willing to invest in our project, we demonstrated that we could match, if not better, some of the voice work happening overseas. It was a runaway success. Today, I believe that more than 50% of GE Capital’s India revenue comes from its call centers.


Aron: How did your experiences at GE lead to the creation of Spectramind?


Roy: We hired Andersen Consulting to help us locate a potential partner who could do business process outsourcing. We discovered at that time that there was no one in the marketplace that was doing BPO processing. American Express ran a captive center [or a proprietary outsourcing operation owned by the company], as did British Airways and Swissair. There were no third-party BPO providers, so GE decided to set up a captive center.


Simultaneously, GE had many business associates who would visit India to look at our IT model. A lot of these Fortune 50 and Fortune 100 firms started coming back and asking us to execute processes for them. That gave us an idea — that there existed a market of people who wanted to utilize the intellectual capital of India, but who did not want to go through the pain of setting up a captive unit there.


Aron: This brings up two issues that are extremely important. First, why is it so difficult for large corporations to set up captive centers? And second, what can a third-party provider offer that a corporation cannot bring to the table using its own resources?


Roy: Wonderful questions. Basically, there are certain aspects of India that the local managers learn to live with (in India). We grow up with the fact that our phones will not work for some part of the time. We grow up with the fact that after you order equipment, there can be a six-month delay before it is delivered. We learn to function in spite of the fact that processing of approvals will not be at lightning speed. We cut our teeth on battling the poor infrastructure and related obstacles and learn to work through and around these problems. Some of the international corporations find it very difficult to get used to all this because it is so radically different from their business environments.


A second aspect is handling the Indian work force, which has its own idiosyncrasies. Indians who work in such jobs are well qualified, highly educated people. They are not a transient workforce marking time while waiting for other things to happen. For them, BPO is a full-time career and they expect some serious thinking on the part of the employers vis-à-vis career paths for them. They also believe that holding on to a job is important, and therefore, what they ask for in a job and in a working environment is very different from the expectations and needs of workers doing similar work in other markets.


These factors, taken together, result in a good deal of complexity, and a lot of the overseas companies don’t necessarily have a deep enough base in India, unlike GE, which had local managers and depth. Sometimes, these companies do not even have an operation in India, so they tend to send someone from headquarters as the trusted person to run it. With no reflection on competencies, this person is often totally at sea; he or she is absolutely unable to figure out what it will take to get the plane to take off. This is not to say that the person won’t succeed eventually, but the cost of setting up a robust, functional unit becomes significantly higher.


Aron: As India starts closing the gap with advanced economies in terms of infrastructure, do you think the pain will decrease, and that large corporations will be able to set up captive centers?


Roy: The pain has decreased over the last eight years, from what I had to do for American Express to what I had to do for Spectramind, but it has not disappeared. There are still significant challenges associated with setting up a service center and getting it off the ground. And it is still far from painless. You know, what they promise is a single-window clearance. Unfortunately, there are 25 single windows you need to go to. That is the reality on the ground, though there are a few exceptions, such as the state of Hyderabad in southern India. Other areas have unbelievable levels of bureaucracy. For instance, Wipro-Spectramind just set up a center in Pune, a city in western India. I counted the number of signatures required on various documents. There were I think 119 places that I signed to get approval to start an operation there.


To be continued: Knowledge at Wharton will publish the second part of this conversation in the next issue.