The need for master data management (MDM) may date back to around 2000 B.C. when a Sumerian accountant used the wrong cuneiform pictogram to record a big sale to an otherwise regular, loyal customer. Upon inspecting the clay tablets later that day, his merchant employer, believing he had attracted a new, deep-pocketed customer, proceeded to order twice as much inventory as he needed.
If our Sumerian merchant had had an easy way to keep all of his customer information in a single place where it could be corrected, compared, and analyzed, he might have recognized and avoided his error.
Even 4000 years later, many enterprises are just beginning to deploy MDM tools that can be used to organize and correct data and help manufacturers catch such mistakes. Although MDM tools have been around for several years, interest among manufacturers has just recently begun to spike. In a survey released in June, Ventana Research found that 49% of all respondents said they have an MDM project planned or under investigation, while 27% have an MDM initiative already under way. Citing the need to improve the management of multiple data entities — with customer, product, and financial data identified as top priorities — more than two out of three companies surveyed stated that a centralized hub is the technical component most critical to their overall MDM strategy.
What's driving the interest in MDM? First, users have begun to grow disappointed with the ability of systems such as customer relationship management (CRM) to keep up with ever-growing data and ever-changing definitions and attributes. Second, passage of the Sarbanes-Oxley Act eliminated the defense of "bad data" in cases of shareowner liability. Under threat of jail, CEOs became legally responsible for the accuracy of all company data, so they're looking for a way to manage data from different sources. And third, sophisticated service-oriented architecture (SOA) tools have begun to emerge that make MDM systems easier to implement.
MDM systems "synchronize and manage data across an enterprise so you can share data across platforms, departments, and domains, and map an understanding of the data across those silos," says Filip Sanna, director of product solutions at Harte-Hanks Trillium Software, a vendor of data quality tools used in MDM systems. "They replace the spaghetti bowl of point-to-point interfaces shared ad hoc with a hub-and-spoke arrangement for sharing and synchronizing data."
An MDM system basically is a central repository where data can be corrected, updated, and then transmitted to data warehouses, supply chain management (SCM), financial and accounting systems, CRM systems, and other applications that support enterprise resource planning. MDM can also manage data provided by suppliers, partners, and customers.
When done right, MDM centralizes all changes and updates to data in real time throughout an entire organization. General Electric, for example, centralizes data management for all its consumer products, says Sunil Gupta, director of solutions marketing at SAP. When there is new data on product models, product upgrades, or service issues — be it in the form of a picture, a model number, or an advisory — the MDM system simultaneously updates sales systems, customer service call-in center systems, and the consumer section of GE Web site.
Enterprises, however, are learning that it is not enough to use MDM only to organize data. It's just as important to clean the data managed by MDM. Incorrect data can cost millions. "Most talk about how data quality is the fundamental
issue," says Karen Shu, principal product manager for Informatica, a supplier of data management systems that recently expanded into data quality by acquiring Similarity Systems.
SAP's Gupta tells of a shoe manufacturer that, after cleaning the redundancies and inaccuracies out of its database of thousands of suppliers, shaved between 1% and 3% off its procurement costs. Noting that the manufacturer typically spent more than $10 billion on parts and assembly, Gupta called that level of saving significant.