Rich Fletcher, director of production operations for commercial lines insurer Crum & Forster, believes goals have to be determined before proceeding on any project. What you have to do first is determine what the ultimate goal is for the transition, he says. Some carriers have the luxury of changing for the sake of change, but most carriers have to point to specific needs. If your goal is to reduce costs and see that in some sort of measurement, you have to look at the project a lot closer, says Fletcher. You have to start taking into consideration overhead costs and everything else to determine what its costing you per policy to do a conversion.

Griffin says he prefers the strategic, top-down approach when considering a conversion. That starts with an evaluation of the entire systems portfolio. Then you take all the business drivers and key indicators of performance, particularly the longer-term strategic plans of the organization, match them up, do a gap analysis, and find out where the major pain points are, he says.

That will give the carrier a road map that makes sense from a business perspective as well as a technology perspective. Then you can apply the normal risk mitigation strategies around that, says Griffin.

{2} the decisionprocess
There are pluses and minuses involved in every deal, and Tom Gravenish, chief of information technology for Equitable Life of Canada, stresses the reasons for change have to be compelling. Many insurers are finding support for a legacy system gradually is disappearing. There is no longer any vendor support for the mainframe system Equit-able is migrating away from, according to Gravenish. The applications we were running limited us in terms of product development.

Gravenish believes there are multiple benefits to be achieved from this project, both on the business side and the IT side. This gives us more responsiveness and more flexibility in terms of product enhancement, he says of the business benefits, adding that time to market should improve for new products and service levels will increase, thanks to more functionality and flexibility.

As for the benefits to IT, Gravenish points to three things: modern technology, a single platform, and better tools. We should be able to deliver more in less time, he says.

Fletcher says Crum & Forster had its data scattered among servers at its regional locations, and by switching to a Windows platform, all data now can be stored in a central database instead of being scattered among the individual servers. It makes any kind of queries that much easier, he says.

Most benefits are tied to the business side, according to Agar. We may see some internal IT productivity, but thats not really the driving force behind our simplification strategy, he says. Its more about better business data and better systems data thats going to be passed between one platform and another.

{3} examine the risk
The length of a legacy conversion project is the biggest challenge facing insurers. Some projects take up to five years or more. Keeping long-term projects in line can be difficult, particularly when members of the project team leave the project or the company. Ideally, project teams would stay together from inception to delivery. GE Res response to this issue has been to create multigenerational plans. If it is a 20-month project, well create deliveries at three, six, nine, 12, 15, and 18 months, says Agar. That way its not building up to one big bang over a 20-month window. The project team needs to feel like it is delivering something early on a project that size.

Another lesson learned throughout the entire GE world is to integrate quickly when mergers or acquisitions take place. The faster you integrate into core systems, the better off you are 12 months after acquisition, says Agar. Clearly, thats a key item if youre in an acquisition mode.

One absolutely has to show significant new business capability in the systems you intend to build or deploy, Griffin adds. Fortunately, insurers under-stand the nature of risk. Insurance companies deal with risk all the time, so they understand long-term risk and short-term risk. Its not an argument thats difficult to make in this kind of business environment that we are operating in.

Affirms Gravenish, This is hardly a trivial or inexpensive exercise.

{4} sell it to the CEO
There are no easy technology sells for a CIO unless a short-cycle return on investment can be shown to the CEO, according to Griffin. CEOs can understand the strategic implications involved in such a project, but make no doubt, It always comes down to dollars and cents, says Griffin. The current economic climate and the fact that some P&C insurers are not making a lot of money right now means the CIO is going to have to show the boss a significant return on investment for the new technology, he asserts.

It takes a special kind of CIO to go to the board of directors and ask for five years and multiple millions of dollars to make a project like this work. Most CIOs wont do that, Griffin says. Theyll tackle it slowly on a year-by-year, quarter-by-quarter budget basis. It takes a real tough, brave CIO to get up there and make the case.

But by avoiding the big push and trying to do things on a piecemeal basis, a company is courting failure, according to Griffin. A project of this scope needs sponsorship from the highest level. IT folkand were all guilty of itare overly optimistic. Most things take twice as long and double the amount of money, he says. So many [CIOs] get burnt because they cant deliver in the time frame they think they have to deliver it in. They think that is the only way the project is going to get approval, so they get unrealistic about what its going to take.

Erb claims it was not a tough sell for ING Re. He says the constraints on business growth and the increasing costs of IT maintenance were obvious to everyone. Still, he believes timing is everything. We are fortunate we started this process back in 1998, so it was really before [the economy] had this recent downturn, he says. If we were to re-look at this project today, it would be a tougher sell.

{5} deciding what comes first
Certain pieces of the puzzle become clear early in a legacy conversion project, but chasing after a quick ROI wont always work. Griffin believes the direction a carrier takes is related to the technology platform the company is about to adopt. If its the first venture into a new technology environment, such as Java J2EE, then I wouldnt recommend the heavy, mission-critical stuff first, he says.

Using claims as an example, Griffin believes the front-end work, such as first notice of loss or first report of injury, should be done at the start. Another possibility is build-ing a Web portal into the system and using that to introduce the new technology into the environment.

Let [IT] learn all the lessons about how to maintain and run that particular technology and then eat into the heavy-lift stuff of the claims machinery as the second or third phase, he says. Companies with a fair amount of internal expertise might be able to tackle bigger chunks, according to Griffin.

Making a thorough examination of what you have and where you want to go is important. Griffin says one of WorldGroups prospects has a relational database schema in a modern relational database environment with some old legacy applications on top of that.

Were going to be building and rolling out some claims systems initially where were actually going to use their existing data structures, he says. Were going to extend them and add a bunch of new tables, but essentially were not going to cut their legacy off from their existing data. Its a strategy to allow them to co-exist.

{6} prepare for a long haul
Its hard not to be afraid when you hear the words five years included in any project discussion, but Griffin reminds carriers le-gacy systems are enormous and enormously complex. If you are a life and health company in a multiline environment, you probably have a hundred years of history, you probably have at least a thousand products, and you are looking at a portfolio of maybe 50,000 to 60,000 programs, he says. You also might have up-wards of 10 million to 15 million lines of code to re-place. There is just no way you can eat that in any sensible fashion and re-place it in a shorter time frame. Thats mission impossible.

{7} be prepared for anything
Erb says he was not prepared for the amount of effort and detail work needed to complete the data transformation and migration portion of the ING Re project. It was huge, he says. We could not stick to the schedules we laid out. It was just too much work. I dont think we realized the extent of the data-cleansing issues, especially when trying to get that data into the new business rules.

Offshore development is an area to examine as well, Agar says, particularly for data conversion, system interfaces, and reporting work. If we need any specialized fields around the software tool we procure, we might outsource pieces of that framework as well, he explains.

Fletcher recommends drawing up a list of benefits and concerns. Put down a list of benefits you think you are going to glean, he says. The concerns list needs to be prioritized. You have to work the concerns list and see where you can go to get those items off that side of the ledger, he suggests.

{8} a lasting legacy
While the future looks bright for companies that have completed the conversion process, the eternal question of the glass being half empty or half full will always remain. Im under the belief there will always be legacy no matter what youre dealing with, says Agar. By the time you do the life cycle of the project, what you implemented has now become legacy.

Nevertheless, insurers that have been through a legacy replacement project are betting that the new system carries them well into the future, and most likely, if the project is successful, someone else will have to worry about the next legacy conversion.

Starting Over
When the legacy mainframe system used by ING Re had reached the end of its usefulness, the life reinsurer was in a quandary. Reinsurance, especially life reinsurance, I would call more of a niche industry from a software perspective, says Larry Erb, CIO for ING Re. There was not a lot of software available, but custom developing a solution was not really a path we wanted to go, says Erb.

Through a sister organization, ING Re became aware of software developer Wyde, which has a policy administration tool that focuses on core insurance technology. While the tool alone could not solve all of ING Res issues, a hybrid conceptpart package solution, part customizationoffered ING Re a beginning. You are starting out with something that has the basic objects code that you needthe functionality to run the business and the ability to customize it to meet your needs, he says. Erb describes the idea of a 100 percent custom solution as very expensive and very risky. Were in the middle, and we think thats a good approach for us.

The first phase of the ING Re project was to increase both the quality and quantity of incoming self-administered data. Typically, the reinsurer would receive 15 to 20 data elements on a session record per client with the old legacy system. We increased that to close to 150 data elements per client that we could use for additional analysis and processing, says Erb.

The life reinsurance standards from ACORD are not as far along as other ACORD standards, but Erb believes they are close enough, so the data is entered through a standard format.

ING Res second step involved the installation of the Wyde tool to replace the administration system. The third phase of the project is one that is just beginningmoving into business intelligence (BI). We recognized if we didnt address the data issue first, BI wouldnt be of much value. Its kind of the method of our overall process.
The biggest challenge in this legacy conversion project, Erb asserts, was the data transformation piece. The reinsurer could have thrown in the towel on the old data when the length of time and effort involved in the process dragged on. The old data was just as important as the new, though. We felt it was important as a reinsurer we have good client information moving forward, he says.

It took roughly 18 months to build the system and close to a year to complete the data transformation. Says Erb: We believe the approach we decided on with this component-based development tool really made the difference with us.

Equitable Life of Canada
www.equitable.ca
$223 million total premium
Whitehill Technologies Inc. Enterprise software
www.whitehilltech.com

crum & forster insurance
(Division of Fairfax Inc.)
www.cfins.com
$779 million net written premium
Insuritys commercial intellisys
www.insurity.com

ge employers re corp
www.geemployersre.com
$2.1 billion in net premium
In the process of implementing a strategic plan to simplify technology environment.

ing re
www.ing-re.com
$795 million in net premiums
SAGE system developed by ING RE using tools from Wyde and CGI.
www.wyde.com
www.cgi.com