A Better Way To Manage Customer Data

IBM, meanwhile, is marketing a set of data governance tools as part of its information integration suite, with modules for managing data quality and identity resolution. But many companies are taking a best-of-breed approach rather than wait for the big vendors to solidify their customer data strategies, Bois says.

People Problem
One useful tool, Data Foundations' OneData, combines business process workflow, change management functionality, and a business rules engine in a way that's fine-tuned for managing customer data. Companies' preferred data models can be imported into OneData, saving businesses from having to adopt new data models and allowing customized implementations in just a couple of days.

But most data governance advocates stress that technology isn't what's holding up progress. Perhaps the biggest obstacle to creating an effective program is getting people to reconsider the notion of data ownership, IBM's Adler says. One of the central characteristics of data governance is the elimination of stovepiped data silos in favor of a more centralized approach in which data stewards make sure records are clean and accurate. That means some data "owners" have to relinquish their fiefdoms--and it's hard to get people to "give up their little turf and work together for a common good," Adler says.

That's why so many businesses continue to function at a level of data management maturity that Aaron Zornes, chief research officer at the Customer Data Integration Institute, describes as "anarchy." Data is coming from multiple sources, getting entered into multiple systems, with no overarching controls to ensure that any one record represents the truth.

Data governance efforts are lagging even in the financial services sector, which not only has the most customer information of any industry but also must contend with the most stringent regulations dictating how that data is managed. In a white paper Zornes is preparing for IBM, he notes that fewer than 30% of financial services companies have centralized their data ownership along a governance model. Worse, two-thirds of those companies haven't documented the related policies and procedures, and three-fourths have no way to evaluate how well it's working.

The goal is to get companies to reach a "federalist" level of maturity in which business and IT work closely together, using service-oriented architectures to create shared data services, and ensuring that data is entered in a consistent manner and managed as a centralized asset, Zornes says. To get there, companies must make a number of cultural and process changes, he says. They have to set up oversight committees to develop policies and procedures that make up the foundation of a governance program. They need to become much more diligent about controlling who has rights to access and make changes to information. Data and metadata must be actively shared across the organization. And in addition to getting data owners to embrace the concept of stewardship--in which they apply their expertise to usher information through a governance program--they have to start adding the position of certified data steward, a job that Zornes expects will become more common in IT departments and business units over the next few years.

Perhaps most important, they have to undertake massive data inventory efforts to identify all their data assets, their value, and the level of risk they present. "If you don't know the value of something, as well as the risks and the costs, how can you manage it?" Zornes asks.

Customer data, however, presents a challenge on that front, says Adler, because ascribing its value is more difficult than computing the worth of other data types, such as patents and software code. "Customer information doesn't have built-in value determinators," he says. "Companies have to evaluate it themselves."

As companies begin to get their arms around this massive challenge, they may find they're able to reduce the number of embarrassing incidents of data exposure. But an even greater motivation is inherent in the bottom-line impact, because bad customer data hits companies where they hurt most: on the bottom line. "What happens if your customer information is 50% wrong?" asks Bell Canada's Garigue. "You're throwing away 50% of your money."