Article: American Express: Performance Under Pressure

Inside Technology Training

January 1999, Volume 3, Number 1, p. 12

"American Express: Performance Under Pressure"

A new performance support system keeps AmEx customers from getting lost in the shuffle. But building it was tougher than Betty Mackay had ever imagined.

by Lauren Gibbons Paul

Prologue

Betty Mackay remembers well the first time she heard about electronic performance support systems (EPSSs). It was 1993. Her training group at Minneapolis-based American Express Financial Advisors (then known as IDS Financial Services) was searching for ways to improve accuracy and reduce processing time in its call centers. Like many call centers, AmEx's were plagued by errors and poor customer service.

Rick Hill, who worked for Mackay, had read about the work of Gloria Gery, a consultant widely credited with creating the EPSS concept. Hill believed an EPSS might help unlock the riddle of poor customer service. Such a system would give call center associates detailed instructions while they were on the phone with American Express customers.

"He thought an EPSS could increase accuracy and greatly reduce or even eliminate the need for training. Since these were major issues in the organization, we were definitely interested," says Mackay, then the training group's director.

Hill, a veteran marathon runner, became an EPSS evangelist, spreading the word to whomever would listen -- both inside the company and out. His enthusiasm was contagious. Mackay was an early convert, although she confesses she wasn't really sure what EPSSs were all about at first. She thought an EPSS sounded like a sophisticated help function. But she soon realized Hill had a lot more in mind.

An EPSS is a computerized application that helps a worker learn and perform a set of tasks. The power of an EPSS is that it removes the need to train the performer on the nuances of a particular job, allowing him to enter a new job without any prior training. The EPSS does the heavy lifting for the worker, boosting him over time to higher levels of performance -- with far less training -- than he would have reached without the system.

Hill was filled with ideas about what an EPSS could do to improve the accuracy of the service associates. For example, rather than making service associates memorize obscure, mainframe-generated codes -- a difficult, error-prone process -- the system could display choices in everyday English and then translate them into mainframe-friendly code behind the scenes. Rather than providing error messages when the worker made a wrong choice, the system could display only valid choices to the performer, thereby preventing the error.

Though she managed training, Mackay was excited at the prospect of using an EPSS to provide bottom-line benefits to the business -- not just to training. An irrepressible woman with wavy dark hair, given to Midwesternisms like "Holy buckets!" and "What a hoot!" Mackay could hardly contain her enthusiasm. She went to Hill's office every evening to pump him for details on the EPSS prototypes he was building.

Early one morning -- a day she'll never forget -- Mackay arrived at her office to the stunning news that Hill, age 40, had died suddenly during his run the previous night. In one moment, she had lost a dear friend -- and the only person at her company who understood EPSSs. Devastated, Mackay vowed to carry on with EPSS in Hill's name. "He left [the EPSS project] to me," she says.

But -- as at all large companies -- management changes and funding cuts meant EPSSs lay dormant in the prototype stage for several years. It wasn't until 1995 that Mackay had her first chance to put Hill's legacy to work.

This is the story of just one of the many EPSSs American Express Financial Advisors (AmEx) has completed to date. It illustrates EPSS's exciting potential, if used in the right circumstances and supported by business leaders. It is also the story of Mackay's struggle to build the system in the face of a near-revolt by the service associates.

The Problem

Bonnie Foley hated her job. As a service associate specializing in estate settlement at American Express Financial Advisors back in 1995, Foley bore the brunt of customer frustration. What's worse, Foley knew the irate customers were right -- she was giving bad service -- but she was helpless to do anything about it. She spent her days fielding a stream of calls from unhappy people.

What led to this shabby service? Simple: the death of an AmEx client. When a client died, that client's heirs wanted to settle his or her accounts, whether they were mutual funds, annuities, or life insurance. Usually, the client owned more than one financial product. This complexity led to all kinds of problems.

In order to initiate a death settlement claim, an AmEx financial advisor -- the person who had provided financial counsel and services to the deceased during his lifetime -- would call the service center where Foley worked and begin the settlement process. Foley handled mutual funds claims. So, when an AmEx mutual fund client died, Foley would begin amassing the complicated paperwork (the death certificate, certificate of ownership, and beneficiary designation) required by federal and state regulations to settle the claim.

The problem was, Foley could only handle paperwork regarding mutual funds. She was powerless to help the AmEx financial advisor or the client's heir with the paperwork for other financial products. Moreover, it would take a year to 18 months for her to settle the mutual fund account. Other AmEx accounts might languish even longer without being settled.

The AmEx service associates didn't talk to each other about the death settlements -- the left hand didn't know what the right hand was doing. The client -- either the AmEx financial advisor or the deceased's heir -- was shuttled from associate to associate. In the end, the client couldn't be sure where the deceased's claims stood as a whole.

"[Heirs and advisors] had a feeling when they sent stuff in that it went into a black hole. They had no clue what was going on," says Betty Mackay, now director of performance support development and transaction re-engineering.

Years might pass before all of the AmEx client's accounts were finally settled. Foley and her colleagues had a name for this poor service. They called it "death by product."

"Clients had to die separately for every [AmEx product] they owned," says Foley. "No one owned the [death settlement] process." Blond, blue-eyed, and energetic, Foley was eager to provide good service, and it hurt that she couldn't explain why the service associates who handled death settlement didn't share case information with each other.

Thankfully for Foley, the situation was ripe for an EPSS, which would help the company give its customers a single point of contact for death settlement claims -- making customers and Foley a whole lot happier.

Folklore Training

Foley's boss, the head of customer service, knew she needed to change the death settlement (called "title transfer") process so that heirs and advisors could deal with a single service associate. The problem was, even in its fragmented state, the title transfer job was essentially untrainable.

Agents who settled estates needed to understand more than 30 different ownership types (e.g., individual, joint, tenants in common, trust or partnership). They needed to know five to seven estate settlement types (mutual fund, annuity, life insurance, and certificate); 50 state estate settlement reporting requirements; 50 state rules regarding treatment of minors and 78 potential required documents -- too much for any one employee to learn. Performing the job was more art than science, and there were great variations in the way people worked.

"There was no way to train it like a regular class," says Mackay.

Service associates had always learned via the "folklore" method: They would ask their neighbors in the next cubicle how to handle each new situation. They prided themselves on memorizing most of the rules in their discrete area. It took nearly two years for a new title transfer associate to master his or her job. While the people in this role, like Foley, had a lower turnover rate than the rest of the service associates, their job was the most complex and demanding in the group. Losing someone presented a crisis.

Potential Solution

Mackay saw an EPSS as the tool that would allow AmEx to reengineer the title transfer job, enabling better customer service and speedier estate settlement. She pitched the idea to the head of the customer service organization.

It wasn't a hard sell, because rather than offering a costly training program, Mackay was offering a solution to a real business problem. "The key is presenting this from the business perspective. Training issues rarely suffice [to justify an EPSS]," says Mackay. "The business problem was getting our clients a single point of contact."

Mackay wanted to embed as many training and support elements into the application as possible, rather than forcing performers to go outside the application for assistance. "You want a lot of support in the context of the task itself," she says. Put enough support in an application and a novice can perform smoothly.
On the other hand, she didn't want to drive employees crazy with too much information. To avoid alienating seasoned employees, Mackay planned to hide a lot of the help behind buttons: Associates would have to click if they wanted a more detailed explanation of how to do a particular action.

Making the Case

By 1995, Mackay had plenty of experience wowing top business executives with the impressive results of two of Rick Hill's early EPSS prototypes. Using Hill's bank authorization prototype, employee accuracy jumped from 89 percent to 97 percent; training time was reduced from 12 hours to two hours; and processing time dropped from a whopping 17 minutes per transaction to just under four minutes.

But Mackay's new application was different, because the goal was customer service improvement -- which is notoriously hard to measure -- as opposed to better accuracy or reduced training time.

Mackay knew they wouldn't see the same impressive results as they got with the prototypes. Each estate is different, and although the goal was to reduce the turnaround time, Mackay knew there was no way to standardize the amount of time it took to settle an estate.

It helped that Mackay wasn't asking for much money compared with other AmEx projects. Even though the system development would have to be done in-house, the project would cost on the order of six figures (she declines to say exactly how much), which was mostly for the developers' salaries. (Subsequent EPSSs have cost AmEx on the order of seven figures for each major project.)

Mackay got the green light to proceed.

Inside Associates' Heads

In June 1995, Mackay and a team consisting of project manager Bill Ricci, a developer, and a select group of service associates (10 were handpicked from the group of 80) began painstakingly deconstructing the tasks that made up the estate settlement process for each of AmEx's different products. EPSS inventor Gloria Gery, principal of Gery Associates, says this stage is notoriously difficult but utterly crucial to an EPSS project. "Performance support requires a deep understanding of the work processes, thinking and business intent for the tasks," says Gery, who lives in Tolland, Mass.

During the first of an endless series of meetings, Mackay realized that the associates thought it was impossible to combine their jobs into one. Nor did they believe a computer application could support a new, merged job function. The associates sat through the meetings in silence with folded arms and averted eyes.

When the associates did speak, Mackay remembers, they voiced grave doubts about the project's viability. They were afraid -- among other things -- of getting stuck with an impossible job. Because they had long been rewarded for their rote-memorization abilities, each thought his or her own area was the most difficult, and that no one else would be able to learn it, even with computerized support.

Mackay asked the associates to suspend their disbelief and help build a prototype. This was difficult because she had no well of credibility on which to draw: None of the associates had ever met her or Ricci before.

"I had to do a big, ‘Trust me.' And of course, they didn't even know me, so that was an interesting problem. I said, ‘Well, let's just see what would happen if we built it," she says.

Mackay brims with self-confidence and enthusiasm -- no one could say she is anything but approachable and down to earth. But as the team got further into the development, she began to worry. To her surprise, she found herself the focal point of the associates' fear, uncertainty, and doubt. "I'm very task-oriented and I don't always pay enough attention to the people side of things. But I probably should have. There was a point along the way when I wish I had done [team building] because it got very tough," she says.

She and Ricci fell into a good-cop/bad-cop routine, with Ricci cast as the good guy. A boyish, lanky redhead, Ricci was a natural to smooth ruffled feathers. On nights before meeting with the associates, Mackay would try out the next day's material on Ricci. He became a good barometer for how the associates would react.

Still, the associates saved their harshest criticisms to unload on their manager and colleagues when they got back to their cubicles.

"This is Minnesota, you know. We're all very polite. We'd never say something nasty to someone's face," says Mackay. "It would have been easier in some ways if they would have, because you couldn't really talk to them about it, you could only hear it third-hand and then try to bring it back into the room and get it out." Mackay found herself flexing muscles she never knew she had.

The early prototype was rudimentary, but it transformed the associates' attitudes. As the prototype did what Mackay had promised, their disbelief fell away. They began to believe job redesign was possible, and even started to get excited about it. Mackay remembers they gave more input during meetings and weren't so hostile.

Then, suddenly, came a crisis.

Anyone Can Do It

As the reality of the job redesign sank in, associates began to wonder whether their jobs were safe. Now, because the system eliminated the need to memorize, virtually anyone could do the job.

"Everyone was very apprehensive. We had to take a good, hard look at why we did things," says Foley.

Working with group manager Carrie Plack, the team identified a set of core competencies for the new title transfer role. These included the ability to interpret information, reason through a difficult problem, give excellent service, and be compassionate. In their new roles, associates would be trusted advisors who would help the heirs of AmEx clients settle their accounts as soon as possible. But not everyone was ready to move into that role.

With 10 years in AmEx customer service management under her belt, Plack was realistic about her employees' willingness to adapt to a reengineered job. "I realized early on that the technology and job design decisions we were making were definitely going to cause employee dissatisfaction," she says.

But Plack believed her organization could not continue to deliver the bad service that made Foley and her colleagues so miserable -- not to mention their clients.

"I believed we needed to build the system. My feeling was ‘We will rehire to cover whatever turnover we had as a result of our decisions, and we will move on,'" Plack says.

The associates grew more and more distraught, and their relationship with Mackay came to a head. Nothing was getting accomplished in the meetings. In October 1995, Mackay felt she had no choice but to temporarily shut down the development process and retrench.

Mackay is not and never was the type to throw in the towel. Daughter of a chamber musician, she played trumpet in an age when that wasn't something girls did. And prior to going into training, she was one of the rare souls who taught eighth-grade English and loved it.

So once again, Mackay turned to Plack for help. She needed to make her understand that the fallout would be severe if they did press on with the project.

"Once I could see that we could do it, technically, it seemed like a crying shame to walk away from it," says Mackay. "What I kept telling [managers] was, ‘Are you willing to put up with the people stuff that's going to come with doing it?' Because they had to be willing to own that. I told them, ‘You're going to have a huge change management situation I can't help you with.'"

Mackay had come to the end of the line. It was up to Plack to make the decision.

Solution

With her boss's blessing, Plack decided to carry on with the project. At Mackay's request she sat down with the associates and briefly stated the scope of the project, sort of an "Associate Bill of Rights" under the new position. For example, Plack and her boss assured the associates they would not get stuck with an impossibly difficult job, that the learning curve for the new job would be less than a year, and that the application would be online in less than a year. This simple act put boundaries around the application and gave the associates the shred of certainty they needed. Slowly, the tide began to turn.

The team was able to complete its development of the prototype, which it rolled out to a group of 14 service associates in August 1996. Immediately, Mackay began getting "buying signals" from the pilot users.

They gave her wish lists of features they wanted to see in the final version, such as letter-writing templates and a personal notepad with which to jot case notes for future reference. After such an inauspicious start, to everyone's surprise, the prototype went comparatively smoothly and was followed by an uneventful rollout to the larger group in January 1997.

Not Everyone Can Do It

The associates fell into the usual breakdown of early adopters, followers, and resisters (the latter constituted about 10 percent of the group). Foley loved the application the first time she used it. "The clients totally appreciated having one point of contact on their death settlement cases," says Foley, who estimates the new Title Transfer system enables her to settle accounts 25 to 50 percent faster, on average.

But most associates were hardly cheering in the aisles. In fact, nearly 40 of the 80 people in death settlement quit before the application was fully deployed. Mackay attributes this to the fact that most people can't assimilate change as quickly as it happens.

"A lot of people think they're going to get an overwhelmingly positive response to a EPSS when it's first introduced, but that isn't always the case. There's so much change, [the performers] don't know how to react," she says.

One early goal of the application was to lower the requirements of the job to entry level so AmEx would be shielded from the pangs of turnover, a fact of life in customer service. And when the application was first rolled out, Plack's group did hire a few people "off the street" and train them as new title transfer associates. But Plack has since backed away from that practice, preferring to have AmEx employees move into the role after they get some general customer service experience. So, at the end of the day, associates' fears that they would be supplanted by a machine turned out to be groundless.

"The system, it's just a tool. It still needs a very smart person who has a really good understanding of our products, and our beneficiaries, and our ownerships, and our disbursement systems. It's one of the more complex jobs in the service organization. So to be really good at it, to deliver the level of service I want my clients getting, I need somebody who has worked their way into that job," says Plack.

Mackay, Plack, and Ricci all say the pain of this EPSS development effort has faded. But they warn that enormous political issues lurk in any EPSS project. Anyone brave -- or foolhardy -- enough to attempt to develop and implement an EPSS must bring these issues to the surface before development begins. Only then can people use the system to change processes, as opposed to merely helping employees do what they were already doing.

Says Mackay, "I was not expecting that degree of resistance. Things are harder than you think on the people side."

Epilogue

Mackay stops her usual frenetic flow of words for a moment and takes a deep breath, reflecting on her EPSS journey. Somehow, through it all -- losing Hill, funding cuts, management changes, and months of hostility from the associates, she managed to keep her faith that the EPSS was the right thing for AmEx customers.

She is a little amazed at what she's been able to achieve. Since Hill's death she's completed three successful EPSS applications, of which Title Transfer was one. She gets a deep satisfaction from taking over Hill's vision and bringing it full circle.

"I owe it all to him. Sometimes, I think he's up there smiling down on what we've done," she says.

Lauren Gibbons Paul is a freelance writer in Belmont, Mass. She can be reached at laurenpaul@sprintmail.com.

What Does an EPSS Look Like?

Most companies consider their EPSSs to be proprietary information. American Express is no different; although Betty Mackay was willing to discuss her application, she declined to share any screens with us. So we went to another expert, Bill Miller, a St. Louis-based design consultant with Edward Jones, a financial services firm.
Miller provided a screen from an EPSS he created. Using this application, clerical staffers in eight aboriginal communities can find and fill in the forms required to register land transactions. The application won a 1996 Award of Excellence from the International Society for Performance Improvement.
According to Miller, the application is based on the idea that work consists of tasks and transactions. He writes, "A task is a distinct unit of work, like filling out a form. Ö A transaction is a series of tasks that must be completed to fulfill an objective."
The screen above is a "guided task selection" screen. It prompts the clerk who's using the system to click on icons in response to questions. In this example, the worker might click on "The deceased" in response to the question on the screen. He'll then be led to another question and series of answers, and so on. Miller writes, "The questions will eventually lead to a checklist of the tasks that are required to complete the transaction."
It is this kind of guided task selection, paired with rich databases, that helps American Express customer service associates respond easily and quickly to their customers' queries.
-- Elaine L. Appleton

An AmEx EPSS at Work

Here's a brief look under the covers of AmEx's Title Transfer application.
In 1995, when Betty Mackay set out to create an electronic performance support system (EPSS) for the title transfer job at AmEx, she had little choice but to develop her own application. Today, there are a few packaged EPSS applications on the market, but you're still likely to have to customize the software.
Mackay's team chose Visual Basic as its programming language. The core of the application -- given the utilitarian name of "Title Transfer"-- is an automated letter-writing system. It links to AmEx's mainframe, from which it gets information about customers and their financial products.
The application is chock-full of support. A string of icons across the top of the screen lets associates know which stage of the title transfer process they're in. For example, "Mr. Yuckface," a green sad-faced icon, signifies the death of a client. The associate clicks on Mr. Yuckface to start the death settlement process. Clicking on an icon depicting a letter takes the performer to a screen that lists all letters generated on the file to date. Red flags indicate documentation that has not yet come in.
Associates estimate that the ability to generate letters automatically saves them about 29 minutes per letter. Since the death settlement associates produce an average of 1,000 letters every month, that's a pretty hefty productivity boost. In the associates' eyes, this is one of the "big wins" of the system, according to project manager Bill Ricci.
Even though EPSS makes life considerably easier for the associates, it's no cinch to build. Mackay and her team had to take all rules for how associates do their jobs and hand code them into the EPSS. When they were finished with the system, the programming documentation was more than three inches thick.