AFLAC: Duck SoupBy Kim S. Nash | Posted 2004-09-01 Email Print
AFLAC's technology staff could not ensure projects would get completed on time and on budget. So it started sniffing out dubious initiatives, before they began.
No one hands you $60 million to spend on technology if you don't know how to plan projects and manage people well enough to make it pay off.
In 2001, AFLAC began to fly right. The company started tracking the time of every person assigned to a project and created a process to sniff out dubious project ideas before they slipped into the development queue.
As the new controls took hold, senior executives allocated a pot of $60 million to be spent over three years on technology infrastructure projects, such as rewriting core mainframe-based underwriting and billing software for Java and Microsoft's .Net systems.
At stake was the credibility of the company's technology staff.
"Most companies don't have good, consistent practices for software development that they use for every project, every time," says David Herron, a principal at David Consulting Group in Medford, N.J., which helps companies improve software development.
And if they don't, reputations suffer. An inability to predict costs or staff needed for an initiative, Herron says, means that "people start not believing you when you talk."
AFLAC had committed to producing much of its software in-house, but often had to wait for the right technologist to tackle projects; a database administrator isn't the best person to upgrade a local area network. Other work would be dropped temporarily as managers moved people with needed skills from one project to another, according to Abeyta.
Predicting costs and timing accurately meant getting a handle on the schedules of all the technology staff. "We need to know where our people are and how long they're going to take, so we can forecast," says Abeyta, a former U.S. Air Force communications and computer systems officer. "Discipline is the word around here."
AFLAC had no central tool to log hours or estimate the human resources needed for upcoming projects. Then, a confluence of business factors spurred the company into action.
AFLAC's revenues dipped 1%, from $9.7 billion in 2000 to $9.6 billion in 2001. Profits of $687 million stayed flat. Yet its name recognition was shooting straight up, thanks to its garrulous waddling duck.
The unexpected success allowed AFLAC to hire 2,300 more agents, boosting its sales force by 22%, and new-premium sales jumped 29% in 2001. Controlling the costs of growth in policy administration and customer service suddenly became a high priority.
AFLAC created a project management office, led by Abeyta and staffed with 20 project managers. He reports to both chief information officer Jim Lester and chief administrative officer Becky Davis. It's a way to ensure each project pursues business goals, not just technical ones.
The Path to Success
The company also instituted new scrutiny with a project management cycle of six phases that it calls "gates.''
Gate meetings are intended to be abrupt stops in the cycle where the company scrutinizes a proposal or project underway. A combination of technology, operations and, ultimately, business unit managers lead the meetings.
Making it through gates 1 and 2 is critical; it's where the most thought and planning happen. About 70% to 75% of the ideas that bubble up from business units get through Gate 1; 90% of those make it through Gate 2 and get built.
To get to Gate 1, a business manager comes to Abeyta's office with an idea. The two work together first to outline the business case, such as cutting redundancy from customer service. Next they rough out a plan to determine the technology resources needed, including any new software and the number of programmers, and what those costs will be.
Labor is usually the most expensive part of any technology project. So figuring out how best to use that $80,000-a-year manager of application services or $60,000 software quality assurance analyst is essential.
To track technology staff, AFLAC installed IT Project Office, a set of project management tools from Primavera Systems in Bala Cynwyd, Pa. The software lets executives and project planners see graphs based on variables such as who is assigned to which projects, how many hours they have put in, how many hours are left and which project they will work on next. The data can be sliced by project, technologist, time frame or business sponsor.
The project management office uses that information every day to run what-if scenarios when researching new proposals. Senior executives such as Davis and Lester use labor data to prioritize business plans. When starting a new project, Lester, for instance, can immediately see who on his staff is available.
If the initial study makes sense, the business manager reaches Gate 1. That is, he presents a project charter to AFLAC's project steering committee of technology and operations managers at its monthly meeting.
Led by Aki Kan, executive vice president of internal operations, the steering committee discusses the charter. The Gate 1 meeting focuses on tangible goals, such as reducing the number of employees who deal with a single customer inquiry. But softer efficiency goals, such as raising customer satisfaction, are also weighed. The committee is in charge of allocating the annual technology budget.
After the group discussion, Abeyta asks for comments or objections. When those have been resolved, he turns to Kan, who makes the call about whether the project continues to Gate 2, the next level of analytical rigor.
If Kan OKs it, he allocates 20% to 30% of the estimated total project cost to fund development of the detailed business case and information-technology impact statement required for Gate 2.
Issues addressed in this phase include: Will the project require a new Microsoft Windows NT server? How many .Net programmers will be needed, and for how many hours? What's the modified internal rate of return? How do the numbers look cast out over five years? The research takes about a month, sometimes longer.
The resulting analysis, which can consume 80 to 100 pages, is summarized in a 10-page document of project goals, costs and payback that is presented to the steering committee at a Gate 2 meeting. A project goal must be succinctly stated and measurable, such as, "Reduce software maintenance by $20,000 by eliminating use of one SQL Server database."
In some cases, two alternatives are presented for the same initiative, such as a proposal this summer to put in a customer relationship management system. One option called for doing all software development in-house, with some help from two or three outside contractors. The other called for going outside for all development and, subsequently, not owning the source code (see "Walking Through a Decision," p. 54).
Getting through Gate 2 is the most critical point in the lifeor deathof a project.
Unless it's a do-or-die regulatory effort, such as systems to ensure compliance with the Sarbanes-Oxley Act, the project must generate concrete monetary benefit, Abeyta says. The more, the better, since all projects are prioritized against each other. A project expected to result in the elimination of three full-time employees in the customer service department after 20 months might not get funded if one with a quicker and better rate of return is also on the board.
"The case has to be tight," he says.
Sometimes technological costs negate business benefit.
Recently AFLAC's procurement department wanted a workflow system so that vendors wanting to sell to the company could submit proposals at a secure Web site. The proposals would be routed electronically to the right purchasing manager, who would make a decision after consulting also electronicallywith appropriate business and accounting staff. The project would have involved modifying existing systems and was deemed too elaborate and time-consuming for the business benefit it would have created, Abeyta says. Instead, AFLAC created a link on its Web site for vendors to offer proposals via e-mail, and the rest of the process is handled manually by the purchasing group.
If approved at Gate 2, projects are rarely cancelled. The idea is to analyze projects painstakingly up front so little work gets done on initiatives that won't pay off.
"AFLAC isn't practicing canceling projects. The maturity is in not starting them," Abeyta says.
Projects that survive the second gate go on to Gate 3 (design), Gate 4 (build and test), Gate 5 (installation) and Gate 6 (benefits assessment), typically within 12 months.
AFLAC recognized its project management shortcomings back in 1999. That's when the company hired consulting firm Gartner to figure out how the technology department spent its days. Gartner found that nearly everyone's time was overallocated. AFLAC was launching too many projects for its staff of less than 400. Plus, the majority of the group's time was spent on keep-the-lights-burning work such as server and e-mail upgrades, rather than creating systems to trigger more sales or make the company more profitable.
"We set goals to move from spending too much time on maintenance to doing more and more time on new project work," says CIO Lester.
In 2002, the first year AFLAC got serious about project management, it completed just nine projects conceived and requested by business units. But with better project management, the numbers have improved, according to Lester. In 2003, 23 new projects were done, including improvements to a sales management application to keep different agents from calling the same potential clients. This year, 40 projects are in the pipeline, with 32 finished so far.
Still, AFLAC must do more to quantify whether finished projects live up to promises, Lester says. Right now, his staff usually checks back 30 days after the project goes live, to make sure the technology works and people are using it. Then the team disbands and individuals move on to new projects.
"Anecdotally, I can say my customers seem happy and I still get invited to lunch," Lester says. "One of the concerns I have, though, is going back and tracking the projects. We don't have a complete picture."
Lester would like the original project team to follow the system for a year or more, to see if it actually met the expected original financial goals. Performing post-mortems further outhe calls it "an I.T. project extended warranty period"would bring back more information on whether goals were met, money was saved and employees were more productive. Lessons could help refine future Gate 1 and Gate 2 proposals, he says.
"We're asking ourselves right now how to provide that type of quality control as projects are delivered," he says, "and not just cut them off at [delivery] saying, 'We're done. See ya.'"
Do that too often, and there won't be any more $60 million pots to spend.
Prioritizing the Work
AFLAC uses software to forecast and schedule work hours of technology staff. This hypothetical example, based on Primavera's Portfolio Analysis module, shows how many hours of work are required from quality assurance managers. This, in turn, warns technology planners to scale back project work or hire more quality assurance managers.
Aflac Base Case
Headquarters: 1932 Wynnton Rd., Columbus, GA 31999
Phone: (706) 323-3431
Business: Supplemental medical and accident insurance
Chief Information Officer: Jim Lester
Financials in 2003: $11.4 billion in revenue; $795 million in profit.
Challenge: To make the most of $60 million in capital earmarked for technology projects since 2002.