According to Jim Davies, director of information systems for the Liability Division of Limit: "Large replacement projects are very risky and unless the risk/ reward ratio is very high, we believe it's better to go for incremental change."
The company has undertaken a series of tactical projects that have allowed its legacy systems to continue to meet the needs of the business, even as it has experienced several structural upheavals.
Limit is a leading provider of capital to the Lloyds of London insurance market. It handles annual premium income of £800 million and employs 750 staff worldwide. Its liability division, which has 125 staff in six locations around the world, looks after the company's non-marine business. This includes underwriting risks such as employers' liability, professional indemnity and public and product liability for large companies.
Limit has viewed new systems with caution, because of the long-term nature of its business. "Claims tend to arise a long time after the risk is undertaken," explains Davies. "For instance, hazards to employees are often not apparent for many years. From an IT point of view, that means we have to keep a lot of historical data so that we can manage claims and track trends underlying our profit-and-loss performance."
With the company's history stretching back to 1974, it has 25 years of open-item historical data, most of which is still used on a regular basis.
Investment decisions are also complicated by the fact that the insurance business is highly cyclical, with rates tending to go up and down on a three-year cycle. According to Davies, that means it's never a good time to invest. "When things are going well, why bother? When they're bad, people don't want to invest." On top of that, the business is seasonal, with most policies being renewed in January and July. Any system implementations have to be carefully planned around these periods of high activity.
Within these constraints, Limit has needed to undertake four major systems projects since 1994 - and in each case, it has plumped for renewal rather than replacement.
The first project involved moving away from systems based on Data General MVS minicomputers, a Data General dialect of Cobol and a proprietary database. "We were seeing poor performance and high maintenance costs, while we had limited access to modern tools and were finding it hard to integrate these systems with PCs," Davies explains.
"Rewriting all the applications would have cost us up to £2m and it would have been difficult to plan the introduction of new systems to fit in with the twice-yearly peak in renewals."
Using tools from Transoft, Limit was able to migrate its existing applications from a proprietary version of Cobol to an open systems version running on Unix against an SQL database. The result, says Davies, was that Limit "avoided around £1.5m in transition costs, was able to preserve its existing business rules and could offer improved performance while cutting maintenance costs - without a major upheaval to the business".
Improved access to its data through SQL also helped Limit during its negotiations with the newly restructured Lloyds insurance market, over how much money Limit would contribute to a fund to bail the market out after the losses of the early 1990s.
"Because we had access to quality information in a timely fashion, we were able to show that our estimates of our reserves were more actuarially responsible than those originally made by Lloyds," Davies explains.
"That helped us negotiate a more realistic contribution to the fund which was several tens of millions less than Lloyds had originally demanded."
The second headache Limit faced was the millennium bug, with the company needing to decide whether to replace or remediate 1.5 million lines of Cobol code. Prompted partly by internal political considerations, a tactical decision was taken to remediate the existing system, again using tools from Transoft.
Davies explains: "We were working against the background of a merger with two other Lloyds syndicates, and there was a desire to move to a single platform for all of them. Choosing a new system just for our division would not have been seen as a good move at that time."
The third challenge was to merge the liability businesses of the three Lloyds syndicates into a single operation. Once agreement from the "names" who comprise the syndicates was obtained, the IT team had just three months to make it happen.
It was not possible to rewrite or implement a new system in that length of time, so Limit used Transoft's middleware to pull data for renewal business from two of the systems - which were developed in Powerhouse and Visual Basic - into the third, Cobol-based system which was used by the largest of the three merging businesses.
Limit was already using component-based technology to wrap its core Cobol functionality for use alongside graphical Visual Basic elements, so this proved relatively straightforward. Work is now ongoing to integrate historical data from the systems.
Finally, Transoft's middleware has been used to improve the service offered to Limit's channel partners, which deal with non-Lloyds brokers. Partners now use a simple front-end developed in Visual Basic to request quotes, which are calculated using the existing functionality within Limit's Cobol system. The same sort of approach has allowed Limit to deliver a Windows interface at low cost on the desktop using thin-client technology.
In short, says Davies, a strategy of incremental renewal and legacy refreshment has allowed Limit "to support a process of continuous improvement and adaptation at minimum risk. I believe we have saved at least £2m in conversion costs and spent less than £400,000 in transition costs.
"We have been able to keep our total IT costs down to less than £10,000 per seat, while achieving a high degree of openness and the option to port to most flavours of Unix and NT," says Davies.
Renewal rather than replacement
The problem: how to support evolving business needs and take advantage of new technologies in an industry averse to big-bang investment and high-risk projects, and whose seasonal nature imposes tight deadlines for implementations
The solution: application of a range of legacy transformation and middleware tools from Transoft
Limit's solution allows it to: