TO TRANSFORM SILOED legacy data into a single "gold" master copy that's usable across the enterprise, insurers such as Aflac are undertaking master data management programs. A systematic, business-driven approach that includes governance ensures MDM success.
It's an issue that has long vexed insurance companies- when information needs to be reconciled, how to extract essential data out of silos without expensive or time-consuming, hand-coded scripting or manual handoffs between departments. In an era when fast-paced growth, overnight market shifts and endless mergers and acquisitions are the norm, insurers can no longer afford to have departments wrestling with multiple copies of incompatible data, or representatives inputting or looking up data multiple times for each transaction. What is needed is a "gold" master copy of the data that is shared across all departments and systems.
The costs of maintaining multiple copies of data across enterprises is hard to justify, says Claudia Imhoff, president of Boulder, Colo.-based data management consultancy Intelligent Solutions Inc. "The consequences of the lack of a master data approach include data redundancy, in which each system, application and department collects their own versions of key business entities," she says. "Duplicate versions of data are stored everywhere. Enterprises spend enormous resources-time, money and people-doing reconciliation because of fractured master data. This reconciliation is a big drain on productivity in the enterprise."
That's because efforts to reconcile diverse and conflicting data sources result in organizational tension. "It leads to arguments over who has the right data, as well as inaccurate reporting and bad decisions," Imhoff adds. "Add to that a lack of confidence in regulatory and financial reports, which leads to fear of audits."
That's why interest is growing in master data management (MDM)-defined as the best practices employed to manage, harmonize and govern the master data associated with an organization's main business entities. MDM is the best recourse to unlock siloed data, often associated with multiple applications for each single line of business. "For many insurance processes, especially first notice of loss, every second counts," says Joshua Schwartz, director at PwC's Diamond Advisory Services, Chicago. "Master data can really help in searching and retrieving existing customer records, or pulling in associated claimants to a party among people, companies, vendors or third parties. This accelerates the processes for adjudicating and processing claims. The time spent by your claims handler or servicing agent to continually repeat and input data not only creates data quality problems downstream, but takes time when servicing customers."
Many insurers have made efforts to address some of these siloed data challenges with enterprise data warehouses or data marts, in which selected historical data is archived within repositories where it is made available for analysis. However, MDM tackles information integration on a higher level than data warehousing. "The purpose of most data warehouses is to analyze historical data, whereas the purpose of MDM is to maintain accurate master data," Imhoff points out. While data in a data warehouse is historical, MDM data is both current as well as possibly historical.
"Many companies have focused on data integration approaches as the means to get answers-whether to share information, present it to servicing agents or to present it to customers," says Schwartz. "But it doesn't give you the insights or the ability to make business-driven decisions. It doesn't give you the opportunity to define who a customer is, or the system of record for a customer e-mail address."
BEYOND THE WAREHOUSE
The evolution from data warehousing to MDM requires a strengthening of data governance practices. At Birmingham, Ala.-based Protective Life Corp., MDM provides a business-driven approach that gets around the limitations of multiple data warehouses, says Mark Underwood, second VP of business intelligence. "We had warehousing from a number of our acquired companies," he says. "We had warehousing as it pertained to specific business initiatives-everything from corporate retention to agency CRM. But we did not want to go through the route of having one major centralized warehouse that's going take seven years to build, and by the time we got there, it would be seven years out of date."
Still, many MDM efforts are springing. At Columbus, Ga.-based Aflac Inc., a metadata repository built within a department is paving the way for an enterprise-wide MDM effort. "This group wouldn't have known all the tenets of master data management, but they had a need to treat some data elements in a fundamentally different and more controlled fashion," says John Keddy, VP of IT application services. "Their approach might be categorized as a 'passive metadata repository approach,' but still provides great value. We are actively reviewing how these processes and overall approach may scale up beyond that department."
As companies engage in MDM approaches-moving toward building a gold master copy that the entire enterprise can reference-lines of communications open up as well. "When you're a 100-year-old insurance company, you've been doing the right thing for the longest time," says Protective's Underwood. "But we had areas that really were not directly aligned, with siloing of our information. We had vertical silos aligned to the products themselves. Annuity data systems stayed with annuities. We had the same problem with the life side. No one necessarily needed to talk to each other; there was no real consensus across the systems. People had to be hands-on with the information."
At Aflac, executives see the movement to an MDM approach in terms of the ability to provide a reliable, enterprise-wide definition of operational metrics that are relevant to the business. "Every metric is beautiful to its mother-but not every metric is key at the strategic level," Keddy says. "We now have a cross-functional group identifying which metrics are truly key-and what MDM processes or operations will apply."
This is where strong data governance can make or break the success of an MDM program. The challenge for MDM efforts is that MDM is a business initiative, versus an IT or data management initiative. Schwartz cites Gartner statistics that predict that 66% of MDM programs over the next four to five years will fail to deliver business value. MDM efforts require an owner for master data assets guided by a data stewardship program. "It's very easy just to fall into a pattern of a huge technology investment in MDM," he cautions. "You have to make up-front investments in process, governance and technology."
GOVERNANCE
Insurers seeing success so far with MDM are taking pains to ensure that business users are intimately involved in MDM efforts. "We must ensure the data, and derivations, are done in a manner that can be consistently produced by the users," Keddy says. "We cannot have situations where lack of data context confuses a user, which could either cause mistakes in analysis or execution. As we identify the key metrics of strategic importance, our group will be proposing a unique level of governance and oversight for those metrics. We must have confidence with those metrics that reflect upon the core of our business and know we can consistently rely on their accuracy. We must provide a platform where we can produce analytics as opposed to endless debates over confidence in data that ultimately get resolved by who produced what."
Data governance is key to Protective's MDM initiatives as well. All MDM activities are addressed through two formal councils that regularly interact to reconcile data management issues. "You can lose a lot of time trying to get two people to reconcile," says Underwood. "We sit down and go through all those data elements that haven't been described, mapped or quantified and qualified. We talk about how we want to manage that data. Some of that process is automated, as part of our actual acquisition of data. And some of it is sitting down and telling businesspeople, 'we understand you have a new product coming out. Help us describe that product as were going to use it from a governance perspective." The company also employed tools to help, including San Francisco-based Embarcadero Technologies Inc.'s ER/Studio.
The governance exercise has generated a great deal of business engagement, Underwood relates. "It's even driven a certain amount of enthusiasm from the business," he says. "We're finally getting the data we want, and we don't have to put all the effort into compiling and sending it out to the other groups in the way that they wanted it. It's all being handled through information services, and everybody's happy."
Getting an MDM effort off the ground requires business participation at a number of levels. To get started, companies need to "create a vision statement and plan for enterprise MDM," says Imhoff. "The vision statement needs to address the business reason for this environment, as well as develop a big-picture roadmap, but also focus on short-term projects." The organization also needs to "create a formal MDM function, with reporting lines of responsibility, and procedures on resolving issues," she adds. Finally, Imhoff says, "incorporate business people into an MDM function. Data stewards, data governance specialists and subject-matter experts need to lead with way, with IT playing a supporting role," she says. "Active and vocal executive support is essential." INN
Joe McKendrick is an author and consultant specializing in IT, based in Doylestown, Pa., and a regular blogger for insurancenetworking.com.
MDM Mythbusting MDM
Many of The popular preconceptions surrounding MDM vary significantly from reality. To clarify some confusing and conflicting points of view on MDM, Gartner Inc. offers this list.
Myth 1: MDM is about implementing a technology.
Reality: MDM is much less about technology and much more about understanding how business processes are supposed to work.
Myth 2: MDM is a project.
Reality: MDM is implemented as a program that forever changes the way the business creates and manages its master data. However, to adopt MDM will require numerous discrete projects.
Myth 3: We don't need MDM; we have an enterprise data warehouse.
Reality: MDM should/will span the organization across all business units and processes (including data stores, operational and analytical).
Myth 4: Implementing ERP means you don't need MDM.
Reality: Enterprise resource planning (ERP) generally means a packaged business application strategy, most often centered on a single, large vendor. ERP, implied but rarely realized for the user organization, is a single process and data model across the organization.
Myth 5: MDM is for large, complex enterprises only.
Reality: The principle of MDM is applied whenever two or more business processes must view or share (master) data. This means that most organizations have a need for the discipline of MDM even if they don't call it that, or if they implement a separate technology called MDM.
Myth 6: Metadata is 'the' key to MDM.
Reality: Metadata is critical to MDM (and many efforts outside MDM), but how metadata is applied in the context of MDM differs by domain, industry, use case and implementation style.
Myth 7: MDM is an IT effort.
Reality: MDM must be driven by the business, a business case, and supported/enabled by IT.
Myth 8: MDM is just too big to do.
Reality: MDM can be and is most presently being adopted one domain or province at a time, and one use case at a time.
Myth 9: MDM is separate to data governance and data quality.
Reality: MDM includes governance (of master data) and data quality (of master data) - MDM cannot be established without them.
Myth 10: MDM technology vendor choice does not matter.
Reality: MDM is complex; rarely do two organizations' MDM programs look alike. Vendor MDM capability has also focused on specialization across data domain, industry, use case, organization and implementation style. Consequently, vendor selection is critical if organizations are to find the right partner.
