To Build or Not to Build
Printer Friendly
Insurance & Technology Magazine - October 2002
Anthony O'Donnell
There was a time when nothing was more natural than to build
large and unique systems, and even now the limitations of the
marketplace may sometimes lead insurance IT executives to think,
"if you want something done right, you might as well do it
yourself." But the DIY attitude is no longer so defensible.
Market pressures are forcing carriers to deliver more creative
offerings much faster than in the past. The question is no longer
simply whether to build or to buy, but how to balance elements
of both when seeking the most economical solution. And, further
complicating the picture, as executives consider the technological
shape of things to come, they also have to consider striking a
new equilibrium in terms of the skill sets of their staffs.
In the past, insurance companies could say, "We're going
to custom-build a system that fits the way we do business,"
according to Judy Johnson, vice president, insurance information
strategies, META Group (Stamford, CT). The industry resisted what
it saw as "cookie-cutter process environments" and "believed
that each company had specific business models, products, needs
and processing environments," she says. Moreover, IT shops
could afford to build huge systems then, Johnson adds. But now
that companies no longer have the luxury of 24-month-and-longer
product cycles, pre-built packages make more sense.
Unfortunately, "companies were often disillusioned with
the vendor options provided, either because they didn't think
the system offered them the functionality they wanted, or they
weren't happy with the services part of the organization,"
Johnson contends. But in large measure vendors of packaged solutions
have fallen short through no greater crime than simply taking
their cue from the insurance industry, and have been more leaders
than followers—which has led to a kind of vicious cycle. "The
vendors have been saying, 'You don't understand what it takes
to be in the software business!'—which is absolutely true—and
the companies are saying, 'You guys don't understand that the
world is changing and you're not providing us with what we need!—which
is also true," Johnson laments.
To bridge this gap, Johnson says, insurance software vendors
have begun to produce more modular offerings, allowing component-based
"best-of-breed" approaches.
The majority of CSC's sales are now of components rather than
traditional "monolithic" full back-office processing
offerings, acknowledges Paul DeFuria, CTO of the vendor's financial
services group. Developing solutions that will satisfy the precise
needs of an insurance company is indeed difficult, he argues,
because vendors can't anticipate the specific difficulties of
dovetailing with the existing functionality of a carrier's systems.
Potential customers "would like to be able to get to this
atomic level of 'I could buy this selection of services as a component,'
but as a practical matter to a software company, it's very hard
to componentize at that level because you have no way of certifying
that your software will deploy and behave properly in production,"
he says.
New Players
As traditional players like CSC were migrating to componentized
approaches, newer companies were moving in. Says META Group's
Johnson: "You saw Castek [Toronto] walk in and start scooping
up market share because they had a component-based system."
Other vendors and consultants have picked up on the trend by offering
development frameworks, such as WorldGroup Consulting's Insureworx,
PwC Consulting's Integrated Financial Services Solution, along
with offerings from British invaders, Sherwood International—which
produces the Amarta business process implementation environment—and
The Innovation Group (TiG), which offers iBOS (innovative business
operating system).
The current success of such solutions is based on the lessons
of large project failures and the time and money it takes to build
from scratch, says Blair Porter, vice president, financial services,
for CGI Group (Toronto), which provides outsourcing as well as
its own framework-type offering, GIOS (Global Insurance Open Solution).
"Some carriers spent a lot of money over the last few years
to get through Y2K and the Internet phase, so there are a lot
of sunk costs they want to be able to harvest," Porter says.
Insurers are trying to maximize investment through re-use of existing
assets, and are looking increasingly to ASP or other hosted offerings.
The question now, according to Porter, is less "Do I buy
or build?" than "How much do I buy versus how much do
I rent?"
All bets are off, however, when the decision can affect a carrier's
ability to differentiate itself from its competitors. That is
the view held by Bob Lukas, senior vice president and CIO of The
Hartford's (Hartford, $167 billion in assets) P&C company.
"The most important criterion to us is whether the function
we're attempting to automate has clear competitive advantage to
The Hartford," he says. In January 2001, Lukas's team began
a project to rebuild its personal lines legacy system by late
2003, "in a way that new product and changes to existing
product can be accommodated much faster than in the legacy environment,"
Lukas adds. "We may buy some little tools or components,
but the overall architecture and capability is something we'll
construct ourselves because of our belief that it will lead us
to competitive advantage."
Other considerations sometimes can make bought solutions a good
choice, Lukas concedes, but generally for "small stand-alone,
commodity-type functions." He has a dimmer view of enterprise
software purchases. "A lot of people, I would say, are sorry
to have made those purchases," he hazards. "We shy away
from big enterprise things unless it's pretty clear that we can
get some capability that will be valuable to us."
Unfortunately, many insurers are not good at caluculating the
cost/benefit of packaged solutions, according to Scott McConnell,
vice president, Cap Gemini Ernst & Young (New York). "Larger
IT shops have traditionally been much more experienced with building
than with buying, and therefore than with the estimates and full-lifecycle
costing associated with buying," he says.
Total Cost of Ownership
Conversely, insurers often underestimate the costs of building
because internal development expenses are not treated the same
as external ones. "You're using existing salary and hardware
expenses, but you're also diverting core resources," says
Jeff Adams, senior vice president for TiG's North American management
consulting division. When insurers decide on internal development,
Adams asks, "Are they really comparing the total cost of,
for example, all the legacy or middleware maintenance they're
going to continue to have to do?"
Carriers are understandably cautious about retooling their legacy
systems but they should look to new and robust ways of looking
at business architecture, Adams advises. "If you're going
to run a process—say, a claims or underwriting process—you need
to understand the series of local or micro events and outcomes
that contribute to an overall workflow," he says. "These
are capabilities that most carriers don't have in abundance."
Some carriers may, however. Craig Lowenthal, CIO, Hartford Financial
Products (HFP), a New York-based subsidiary of The Hartford, agrees
that buying is an attractive option in many cases, but says "if
you buy it, try not to modify it." If HFP's legacy issuance
system were broken, Lowenthal says by way of example, the company
might have replaced it with a bought solution. However, since
that system served the company's needs well, HFP kept it and modified
the front end to align with business processes and workflow. "It
could be a big mistake for me to buy a new system. Sure, I get
a brand-new front end, XML and potentially other benefits, but
then it takes me multiple years to rebuild the critical back-end
nuts and bolts of the system," Lowenthal says. Gradual change
has resulted in proven processes over the years, allowing HFP's
IT department to "leverage the known, while continually upgrading
the user experience," he adds.
This summer HFP took a combined buy/build approach toward an
electronic filing and document imaging requirement. A natural
"buy" option would have been to bring in a whole application
that managed document retrieval and storage and required certain
hardware and infrastructure investments, Lowenthal suggests. With
such an approach "you're getting into a huge cost proposition,
and on top of that, you're radically changing users' workflow."
HFP's solution was to buy a combination of eCopy (Nashua, NH)
scanner add-ons to existing digital copiers and $150 scanner attachments
for existing employee laser printers. "They can go to the
copier and scan, or they can do it right at their desk,"
Lowenthal says. Since HFP's customer correspondence is created
in Word, "using Windows APIs, we built a process whereby
instead of hitting the save button, employees hit a custom button
that will print/fax/e-mail (customer's choice) the document, as
well as save it to our network document store. Then, as part of
our existing underwriter workstation, the employee types the account
name and the system brings up all the documents associated with
it."
The Costs of Innovation
However, even the most innovative approaches to building can
sometimes end up being very costly. For example,before it became
OneBeacon (Boston, $3.9 billion 2001 earned premium), CGU Insurance
attempted a tool-kit/component approach to the replacement of
legacy systems that had suffered from a lack of maintenance investment.
Since the IT department had not profited skills-wise from adquate
maintenance efforts, such an ambitious project was like "going
from 0 to 60," says Mike Natan, CIO, One Beacon. "They
went with some very cutting-edge technologies whose implications
they hadn't really thought through," he says. As the solution
took shape, it became clear that it was going to be fairly unmanageable,
Natan relates.
After White Mountains Insurance Group (Hamilton, Bermuda, $16.2
billion in assets) bought CGU, Natan was brought in as consultant
in June 2001. Within a month he decided to kill the project—which
to that point had consumed about $75 million.
Natan investigated a number of market options to start anew,
mostly "buy-and-integrate" solutions in the $25 million
range. The solution chosen after a two-month evaluation process
was to resurrect an insurance framework built by General Accident
(which had merged with Commercial Union to form CGU) using a Sapiens
(Rehovot, Israel) tool kit. "To resurrect that, upgrade it,
bring it to a current state, add functionality and interface it
to the current systems turned out to be the best option for us,"
Natan says—and at a total cost of about $11 million.
Building can even yield marketable results, as it did at Medical
Mutual of Ohio (MMOH, Cleveland), which spun off its Westlake,
OH-based IT operations to form Antares Management Solutions. "We
felt there was a market for companies that are good at insurance
but can't afford mistakes," says Paul Apostle, Antares vice
president, IS.
Antares offers a "third way" in the build-vs.-buy
debate—the "rent" option. Antares specializes in providing
business process and technology outsourcing for healthcare and
related industries. "The biggest issue in finding a solution
is often urgency—which rules out a build," Apostle says.
"The whole 'con' on buying is it winds up with the longest
lead time getting to product and it's the most expensive option
of the three we talk about."
Renting, by contrast, offers the shortest lead time, along with
other advantages. "If you buy a package, you still have to
reinvent the wheel yourself," Apostle argues. "If you
rent, you're probably working with someone who's done it a few
times and knows the pitfalls." But perhaps a greater benefit
is the normalization of costs. "Someone walking into a 'build'
has to be prepared for scope creep and delayed delivery times,"
says Apostle.
Refining Skillsets
Farmer's Insurance (Los Angeles, $11.5 billion in total assets)
has a large IT shop proud of its building abilities, but it is
leveraging the rent option to sharpen its competitive edge. At
Farmer's, that means not only reducing the costs of IT operations,
but refining the skills of IT staff. The carrier structures IT
into two tiers, according to James Fridenberg, vice president,
applications development. The top tier is all our architects,
analysts, subject matter experts and designers, the people who
have core knowledge of our systems and applications. Those are
the people we want to groom and invest in," he says. "The
bottom tier is where we look at outsourcing opportunities—the
heads-down programer/coder." Within that concept, Farmers
maximizes its outsourcing of application maintenance, to such
partners as Wipro (Bangalore) and Cognizant Technology Solutions,
a Teaneck, NJ-based firm whose software development centers are
located in India.
Build to Sell?
Faced with a demand to produce an application for AIG's (New
York, $268 billion in assets) loss control department, Leon Busiello,
director, management information systems, AIG Consultants, was
driven to build it by a typical industry problem. "When we
looked for a system that did what we needed it to do, we didn't
find anything we liked," he recalls. "The basic work
management program we found wasn't made for loss control, and
if we bought it we'd probably have to spend as much money customizing
it."
Busiello describes the existing solution as an extremely inflexible
"half-paper, half-mainframe" application that required
massive efforts when changes in report structure were needed.
The solution was notoriously user-unfriendly, and produced only
hard-copy output that required physical filing accommodations
off-site.
"We wanted something that was centrally based, could give
real-time information and would serve as an archive, as well as
a tracking and order system," Busiello explains. "So
we built something accessible to our underwriting groups so they
could go online and order loss control directly from their desks."
The system, called COATS (consultant order and tracking system),
was built as a client/server application with a VisualBasic front
end in early 1998. About a year-and-a-half later the system was
Internet-enabled.
The efficiencies that the solution delivered resulted in its
becoming "more of our accounting system—it keeps morphing
into other operations," Busiello says. It eventually attracted
the attention of Global Energy, an engineering group within AIG.
"They came over and wanted to 'rob' some of the code, so
we actually sold it to them," in the spring of 2002. The
application was later rolled out to American International Underwriters
Engineering, AIG's international engineering group (which was
in the process of being merged into AIG Consultants), Busiello
says. "We've already started with the United Kingdom and
Europe."
This article orginally appeared in the October 2002 issue
of Insurance and Technology
Magazine. Reprinted by Permission.