Piece Plan

"But there is a third choice, and that is to go it alone and build your own system," says Jenkins. "That's been my druthers for the last 15 years. I rather would have control over my own destiny than leave it to someone else. Our principle is to reuse before we buy, and buy before we build," he continues. "If there is a product that meets your requirements and you can plug and play, you would be better off to do that. It's cheaper in the long run.

"Despite that policy, we have decided, based on nothing currently available we need, that we will build our own system. I went through the same wars 15 years ago with General Accident Insurance and reached the same conclusion," he says.

"We've done a blended approach that involves mature components tied together with SOA," Jenkins explains. The blend includes a rules engine from Pegasystems for upfront underwriting automation as well as a best-of-breed rating engine (Skywire Software's InsBridge).

"We also have a workflow engine from ImageRight," he says. "We call this a construct philosophy–buying pre-fabs and tying them together, then we build the rest."

So far, the combined components are working "fairly well," he asserts. "We're about to go into systems testing."

Expanding on his company's requirements in terms of policy administration, Jenkins notes Penn National "needs to have a system in place that gives us agility and flexibility and that allows us to keep up with business needs. Business needs can change over weeks or months," he continues. "It takes 18 months to catch up tech-wise using older systems. IT is in place to support the business, and if you can't support the business, you're a problem. So, we're looking for agility."

According to Jenkins, the component approach gives Penn National that kind of flexibility. "Changes take a lot less than 18 months with agile systems," he points out. "We're looking at weeks, not months."

Another key issue, he says, is accessibility of data. "It can be difficult to get at data with older systems," he indicates. "We want to make sure we have a standard data model, a standard business glossary, so we don't need translating interfaces.

"The industry doesn't have a standard data dictionary or model at this point, so we have to establish our own standards," he says. "Some data is embedded in code and put in different silos in systems, or systems may aggregate code, so it is difficult to break out in order to do predictive analytics. Companies run on data. The winners make good decisions based on data they have," he adds.

In terms of a wish list for policy administration systems, Jenkins says, "Nirvana would be one holistic system with commercial, personal, billing, and claims. In the center is the data; it's data-centric. Each system uses the same data."

It also would be important the system has a common look and feel, utilizing the same screens and business rules across lines of business, he says. "We should be able to build in knowledge and automate mundane tasks–the STP concept."

Will such a product system ever be developed? "That's our vision here," states Jenkins. "It is possible. The vendors are getting a lot better, and some are moving in this direction, and I think over time they will be able to do this."

NEVER ASSUME

When it comes to policy administration systems, "you make assumptions about what you'll find out there," states Craig Lowenthal, senior vice president and CIO of New York Marine and General Insurance Company (NYMAGIC). "I assumed wrong."

According to Lowenthal, "almost every vendor has a different definition of policy administration. For some, it is strictly policy entry, while others include issuance, billing, and/or reinsurance."

In addition, "some vendors segment into different size carriers they're looking at," he notes. "For a carrier with diversified lines, [vendors] make an assumption the carrier has a lot of components and is not looking to do a wholesale replacement. A vendor like that is not good for my current needs.

"Based on our business model and strategy, I need a holistic approach. I don't want to stitch together products in a best-of-breed approach," he says. "You can make an argument for best of breed. Yes, each module may do a better job, but each module typically has its own database. You have to bring [those databases] together with SOA or some other technology. So, I want a vendor with soup to nuts, maybe even with general ledger accounting thrown in," he continues. "I'm looking for submission and policy entry, billing, entry, issuance, claims, and reinsurance."

Lowenthal says his company has found a product that covers all those bases, "but it also allows me to decouple modules. I may not need the best-of-breed [module], but now I can couple it in if I want to do it later." With such build-or-buy flexibility, "there's a ton of advantages either way. Now, I can have my cake and eat it, too," he states. "There's no competitive advantage to building a policy administration system from scratch."

The key point is to buy a system that is flexible enough for business changes and has entry and exit points at every screen and process, he explains. "That lets me write in functionality I develop that can do what the [base] system could not do well itself.

"I looked at Celent's report of top policy administration system vendors, then I brought several vendors in, and they all had different pros and cons, which makes sense," he remarks. "But I found only one system that offers this turnkey solution, from ISI, based in Toronto. Over the last four years [ISI] created a new system using latest technologies, and it works very well." Other companies, he says, updated by replacing modules with new technology or acquiring a system from another vendor, "but then you have a hodgepodge of look and feel and different technologies," he adds. "It's not intuitive at all."

Lowenthal admits to being surprised many policy administration system vendors are in this situation. "There is a need in the market for [systems designed for] small to midsize insurance carriers," he contends. "A lot of them don't have big IT departments, but really no vendor out there caters to that. Products tend to be great at one thing but not others."

For an ideal policy administration system, Lowenthal indicates he would like to see "a lot more integration with reinsurance. Most systems either ignore or do a basic-to-poor job with this, but many companies need this ability. Many are using batch processing to older reinsurance applications that require overnight processing, and that is ridiculous in this day and age."

A NEW WORLD

Dede DeRosa, CIO of business insurance and financial professional insurance at Travelers, cites a shift in thinking about policy administration systems.

"In the old days, every five to 10 years we would do an evaluation of our core policy administration platform and move to next technology," she explains. "We had to decide whether to build or buy. That kind of simplistic strategy isn't viable anymore, with the explosion of distribution channels and mergers and acquisitions.

"There are so many options available now. The policy administration space is not as narrow as we used to think about it. In many cases, we are looking for end-to-end solutions," she notes.

"One of the things that is nice is there is a lot more ability to plug and play with different solutions," she continues. "We don't have to look at total replacement to move forward to better technology. We can focus on bottlenecks and address them."

Still, there are bound to be gaps, she points out. "Despite being highly regulated, we have our own unique ways of doing things, so it's hard to get out-of-the-box software to fit what we are doing," she states. "It's a complex world we live in. It would be nice to go somewhere and find a comprehensive solution we can apply to our business model, but it's not there. So, we're looking for a company that can look at pain points and bottlenecks and address them in a component fashion. That gives us more options to mix and match," she says.

It seems the vendor-customer relationship also has changed, DeRosa adds. "In prior lives I have looked at actual functionality of system. Now, we're assessing quality and caliber of talent in that firm as much as the software it has. We are looking to get up and running quickly; looking for a partnership with the company," she continues. "How tech savvy is the vendor? How business savvy is it? A lot of subjective or soft qualities are at least as important as the actual product itself these days."

According to DeRosa, "finding good companies is as important as finding good products. Products aren't static; they're always moving forward to keep pace with technology advances." She cites a good customer-vendor relationship as "advantageous for the vendors and for us."

What would she wish for in a policy administration system? "I do think one thing we really are looking for is software that is able to be configured and changed by our business users–not IT people," she says. "IT's role is to get software in-house and establish a baseline but then turn over a lot of the maintenance and change to the users. Not having to come to IT to change a rate is a great thing, so the more we can make that a business function, the better off we are."

She adds mix-and-match components as well as open architecture to allow integration with other software also are important. "Everybody is moving away from proprietary packages to component-based systems," she says.