Semiconductor giant Intel Corp. can sum up its internal master data management (MDM) strategy in four simple words:...
Standardize, consolidate, optimize and utilize.
Those four words represent the Santa Clara, Calif.-based microprocessor manufacturer's "silver bullet" for achieving success with MDM, according to Brian Stinson, Intel's enterprise data product manager, who spoke earlier this month at IT research firm Gartner Inc.'s annual MDM Summit in Los Angeles.
What is master data management (MDM)?
Gartner Inc. defines master data management (MDM) as a technology-enabled discipline in which the business and IT sides of an enterprise work together to ensure the uniformity, accuracy, stewardship, semantic consistency and accountability of shared master data sets.
Stinson, who had several pieces of advice for organizations seeking to execute an MDM project plan, said the first phases -- standardization and consolidation -- may lack appeal, but they are certainly worth it in the long term.
The standardization process involves meeting with representatives from all business units and reaching an agreement on data governance rules, such as the precise spellings and definitions of commonly used terms related to customers, products, suppliers or any other key business entities that need to be mastered.
"I can't overemphasize the value of doing standards and data governance. Having this kind of agreement on the items that we buy and items that we are going to market or sell is so important," Stinson said. "It's not sexy. It's not easy. But it really does pay dividends in the end."
The next not-so-sexy step -- consolidation -- focuses on creating a hub for approved terms, definitions and records. For example, Intel put highly reliable customer records into a hub that can be accessed and used by sales, marketing and other business applications.
Once the standardization and consolidation work is complete "you can go on to the sexy stuff," which includes the optimization and utilization phases, Stinson said. The optimization phase focuses on building out MDM workflows and adding matching, merging and cleansing capabilities.
"If you have standardized and you have consolidated and you've got your systems of record and your record of origin, then you can start adding these capabilities," Stinson said.
Once that is complete, it's time to start using the data. Problems are likely to arise during this phase, Stinson warned, but that's a good thing and all part of the MDM refinement process.
"When people start using the data, what do people find? They find problems, or they find that they wanted something different, or they find they wanted something more, and the cycle repeats," Stinson explained.
After explaining the four parts of Intel's MDM strategy, Stinson recommended that organizations work to establish metrics for measuring the benefits of MDM activities and also for capturing the consequences of bad data.
"What keeps the cycle going is governance," Stinson said. "Once you get some momentum going, it gets easier."
For more on master data management strategy
Get advice from Gartner Inc. on how to reach the next level of MDM maturity
Learn why office politics and MDM can be a dangerous mix
Get Gartner’s take on the connection between MDM and “big data”
Intel focuses MDM strategy on six key domains
Intel's MDM journey began around 2005-2006, when it began to enter new markets in an effort to become a "platform company," Stinson said. Some of the changes happening at the time included new investments designed to increase supply chain efficiencies and an effort to improve customer satisfaction.
During the same time period, Intel's enterprise resource planning (ERP) system started to show its age, Stinson said, and the company decided it was time for an upgrade. Intel also decided that if it wanted to do things right, the ERP upgrade would have to be accompanied by an all-out MDM initiative.
"The domains that we decided we had to master included customer, supplier, item, worker, location and commodity," Stinson said. "These are the six areas that we had to get right for our ERP."
Multiple MDM tools may be required
Stinson said he learned some key lessons during Intel's ongoing MDM efforts. One thing he learned is that while it's important to standardize business terms and definitions, standardizing on one MDM technology across multiple domains may not always be possible, regardless of what vendors claim.
Intel began the process of mastering the six key domains by using the MDM tools that came standard with the company's new ERP system. But those tools only worked well in a handful of domains.
"What we had was a solution that was really good at cataloging, so we ended up using it for our location data, our commodity data, our supplier catalog," Stinson said. "Then we attempted to use it for our item or product data, but we struggled with it a little bit and ended up with kind of a custom hybrid solution."
While Stinson declined to name any of the software vendors that Intel is using for MDM, he said the company ultimately ended up buying two more sets of MDM tools that would help the company master the supplier and customer domains.
"We had a standard approach," Stinson said. "But when it comes to actually implementing MDM, you need to tailor your solution to the domain and its requirements."