Smart CIOs are experimenting with new Web-based technologies to integrate their customer data applications without having to rip out their legacy systems. But before they plunge into the implementation, they need to craft a data management strategy

Reader ROI

• Data management strategies for CRM success
• Why business ownership of customer data is so critical
• The role Web services and service-oriented architectures play in new CRM approaches

The multiple mergers that formed insurer UnumProvident in the late 90s aggregated billions in revenue, assembled thousands of employees - and created a quagmire of customer data systems that couldn't talk to each other. In all, between Provident, Colonial, Paul Revere and Unum there were 34 disconnected policy and claims back-office systems, all loaded with critical customer data. As a result, "it was very difficult to get your hands around the information", understates Bob Dolmovich, UnumProvident's VP of business integration and data architecture. One UnumProvident customer's account, for instance, might exist in multiple places within the newly combined company, leading, of course, to a great deal of waste.

For the first couple of years after the mergers, UnumProvident used a home-grown data-store solution as a Band-Aid. But by 2004 the $US10 billion disability insurer felt compelled to embark on a new master data management strategy aimed at uniting the company's disparate pockets of customer data, including account activity, premiums and payments. Core to UnumProvident's strategy would be a customer data integration (CDI) hub, built on service-oriented architecture (SOA), using a standard set of protocols for connecting applications via the Web (in effect, Web services). The project, begun in early 2005, has already improved data quality, soothed the multiple customer records headaches and created the possibility for a companywide, in-depth customer analysis. But as Dolmovich acknowledges, there's still a long way to go. Of those original 34 systems, he has been able to get rid of only four to date. But he's still optimistic.

"The desired end state is a CDI hub that has information about all customers across all products," he says.

The Quest for the CRM Holy Grail

Despite the long, slow slog, Dolmovich is hoping that the new CDI approach will ultimately give his company the 360-degree view of the customer that has been promised by vendors since the dawn of CRM. In the late 90s, enterprise software vendors like Oracle, PeopleSoft and Siebel sold the single-customer view as CRM's holy grail. But implementation flameouts and legacy integration nightmares soured many CIOs on these expensive enterprise-wide roll-outs. More recently, on-demand CRM has generated a lot of buzz, but it too has run into scaling and integration problems, particularly at large companies (see "The Truth About On-Demand CRM", CIO March).

A CDI hub differs from a traditional CRM solution in that a CDI hub allows a company to automatically integrate all of its customer data into one database, while ensuring the quality and accuracy of the data before it is sent to the hub's central store for safekeeping. A stand-alone CRM system can't do that because it can't be integrated with the billing, marketing, ERP and supply chain systems that house customer data, and it has no way to address inconsistent data across platforms.

What is also missing in many of these earlier CRM implementations, experts say, is a management strategy that identifies important customer data and lays out a disciplined governance process to ensure its quality and its integration with critical systems. "Unless companies have a broad strategy about how [to manage their data], no matter how good transactional systems are, they're not going to be able to deliver," says Ronda Krier, Oracle's senior director of product strategy.

An increasing number of CIOs are now realizing the importance of such a data management strategy and are experimenting with Web services technology to unite legacy systems with new applications without having to rip and replace everything. Many of these CIOs are building a service-oriented architecture that can integrate their divergent applications into a CDI hub via the Web.

However, much like the CRM implementations that preceded it, this new approach is neither cheap nor fast. Ray Wang, Forrester Research's principal analyst of enterprise applications, says that average CDI installations cost nearly $US5 million for licences and implementation services. And they can take much longer than expected. (UnumProvident's CDI implementation, still unfinished, has taken a year so far.) But that's still cheaper and quicker than ripping out all of a company's old systems and installing proprietary enterprise CRM.