Since the days when the first computer spit out data from its miles of wires and labyrinths of vacuum tubes, the euphoria ("Be gone, sliderule!") has been replaced by a feeling of dread. What happened when you plugged in the next UNIVAC in an adjacent building? How could you reconcile the series of 1s and 0s from one computer when you had another stream of data coming from the next one?
This has been a challenge for more than 60 years: How to reconcile multiple views of customers, products, assets, inventory, and any other data elements. With the cost of data storage having declined over the years and the power of computers having increased, the problem of disconnected data is now at a crisis level in many organizations.
Barely a week goes by without a story that ultimately involves a data management problem. A healthcare company gets fined for processing fraudulent transactions. A bank wakes up to a public relations nightmare by foreclosing on a house that wasn't actually eligible for foreclosure. These are all problems caused by multiple, disparate views of data.
In the past decade, the de riguer technology for every information quality problem has been master data management (MDM). Vendors (and full disclosure, this writer is as guilty as anyone else) have presented MDM as the cure-all for all data-driven ills.
Unfortunately, the hype has smothered a dirty little truth. The issues with having "multiple versions of the truth" were supposed to be reconciled by data warehouses, ERP systems, and customer relationship management (CRM) already. MDM takes that abstraction to a cross-functional level, but the general focus is the same: To create a single, unified view of a data entity.
As we continue through the first quarter of 2011, there is notable weariness in the general market. At a recent tradeshow, several attendees rolled their eyes at the mere mention of MDM. One disenchanted IT veteran asked, "If it's so great, where are the success stories? Where is the nirvana?"
Gartner astutely pointed out in its "Hype Cycle for Master Data Management, 2010" report that most of the early forms of MDM (MDM for customer data and MDM for product data) are reaching the "Trough of Disillusionment." As the MDM market enters a phase of considerable skepticism, how does one go forward with these programs? The answer lies in focusing on what MDM is, not on what it's perceived to be.
1) Figure out if MDM is what you need or if another initiative will suffice. If your data management challenges are more finite or if you have a smaller number of applications within the organization, MDM might be overkill for your situation. There are other ways to achieve a "single view" outside of MDM. Creating a reference data "lookup" or migrating other data to an existing CRM or ERP system can achieve many of the goals of MDM without the costs.
2) Realize that MDM is not a technology but a shift in mindset. After embarking on an MDM deployment, people soon understand that technology is not the most difficult element. It's the countless meetings, often accompanied by turf wars and political skullduggery that results from unifying systems. Seemingly easy questions like "How do we define a customer?" can be subject to a dozen different interpretations based on one's position in an organization. Recognize at the outset that MDM is a marathon, and not a sprint.
3) With this in mind, structure your MDM program to deliver results as soon as possible. Very few executives have the time or the patience to give a project five years to pay off. The good news is that MDM projects can yield some immediate results. Having a sound data governance program in place is critical to move forward in an MDM deployment, and creating one will involve establishing the business rules necessary to address entity definitions. These rules can help an organization refine processes and mitigate risks in short order.
4) Finally, don't try to boil the ocean, or whatever metaphor applies here. In the same spirit as the previous point, it's imperative that you start an MDM program with an eye towards a more immediate value. For example, a company that does pharmaceutical clinical trials realized that its data pain was the inability to review results across projects, physicians, and other variables. For three years, the company worked at the application level to install and manage the business rules identified in their data governance program. When it was time to move to a true MDM system, the backbone was already in place and everything went smoother than expected.
Daniel Teachey is senior director of marketing at DataFlux, a software provider of data management technology and services.