Many of today's cost-conscious food shoppers buy store brands -- like Costco's Kirkland or Stop & Shop's Nature's Promise -- rather than national brands. Industry figures show that nearly 1 out of 4 products bought in U.S. supermarkets last year was a private-label brand.
For Daymon Worldwide Inc., which helps retailers market private brands, the industry boom required a major overhaul of IT and business processes to support 5,000-plus suppliers, over 120 retailers and 500,000 products. "We've had to quickly change to stay ahead of the market and keep up with growth," says Abhishak Beniwal, senior director of IT.
A key step was to get control of the company's sprawling collection of product and supplier data, by using a set of processes and technologies called master data management (MDM).
Case Study: Health Care Service Corp.
Implementers are quick to say that master data management initiatives are only as good as the processes and people surrounding them. To that end, Rick Biederstadt, divisional vice president of enterprise information strategy and management at Health Care Service Corp. (HCSC), took steps to ensure that momentum would continue on the organization's MDM initiative.
The health insurer turned to MDM primarily because it wanted to start viewing its 12.5 million members as individuals, rather than as a big collection of members, and to lower health care costs by supporting each individual's wellness. That goal required creating a unique identifier for each member and collecting profile data in one place, to enable a 360-degree view of each person's interactions with the health care system, Biederstadt says.
He chose MDM technology from IBM for the project, which began in early 2009. By July 2009, HCSC had completed its first data load and established governance processes. By late 2009, the MDM system had gone live and, according to Biederstadt, was generating business benefits.
For instance, obtaining a single view of a member led to better retention and improved customer satisfaction scores. The improvements in data management also reduced the organization's exposure to regulatory compliance penalties.
These successes would have been impossible, Biederstadt says, if the overall culture at HCSC hadn't also undergone a transformation: As part of the MDM effort, employees were encouraged to embrace accurate customer data as an important corporate goal.
The key was changing the mind-set of employees about who owns the data, a change that requires executive sponsorship, careful governance and continual communication. HCSC even had cards printed with a capital "E," for "Enterprise," and asked employees to affix them to the back of their ID badges. This was meant to remind them that data is an enterprise asset, Biederstadt says.
"When people started to think locally or divisionally about data, we asked them to turn their badge around," he explains. "Anyone could hold up their E card and start thinking of data belonging to the enterprise again."
Before Daymon began its MDM initiative last year, its product and supplier information was maintained by numerous people in 200 offices around the world, and each change had to be made directly in every line-of-business application.
Today, one centralized group validates the data, which flows to an MDM system from Kalido Ltd. The data is stored in an enterprise data warehouse and is routed to the appropriate line-of-business applications. Invalid or incomplete data is routed to the right individual in the business who can correct it, to maintain data quality.
Now, when a new product is introduced or new packaging is requested, the change is "a simple business exercise" that can be accomplished without touching operational systems, Beniwal says. Furthermore, the cost of managing supplier information has decreased, and Daymon can move into the types of advanced analytics that will enable it to capitalize on future growth opportunities in the private brand business, he says.
So, What Is MDM?
As companies such as Daymon tackle today's business challenges, many are facing the hard truth that they need to go back to the mountains of poorly managed customer, product, supplier and employee data that has accumulated over the years and make some sense of it. That's where MDM comes in.
In a recent survey of 131 companies by analyst firm Information Difference Ltd., 42% of respondents said they had implemented or were in the process of implementing an MDM project. Nearly one-third reported having deployed two or more MDM programs. Just 22% said they had no plans to implement MDM.
But despite their popularity, MDM projects are also renowned for billowing out of scope, exhausting budgets and running out of gas. Often, the very people who would benefit from them see little connection between the hard work involved in making MDM succeed and the results these new systems and processes are supposed to yield.
And no wonder: Enterprise MDM initiatives cost in the millions of dollars for most companies, Beniwal says. They require months, if not years, of organized and persistent effort on the part of cross-functional teams and should really be considered more of a journey than a project with an actual end.
For that reason, no one should begin an MDM effort without support from the very top of the organization, defined metrics of success and a very specific vision of the business goals.
Trying to define MDM tends to trigger arguments among vendors and analysts, but Gartner Inc. analyst Bill Swanton says it can be explained simply. "MDM is a discipline and a process for keeping data accurate and consistent enough so that your applications perform the way you need them to in order to run the business," he says.
Beniwal emphasizes that MDM is a business process for managing information about core business entities such as "customer," "supplier," "products" and "employee." He maintains that MDM combines people, processes and technology to create and maintain a clean, consistent "single version of the truth" that can be used by all business applications within the organization.
While MDM is all too often seen as a technology, a general rule of thumb is that an MDM project is actually 20% technology, 80% new processes.
In fact, at Daymon, Beniwal avoids using the three-letter acronym altogether. "If we say 'MDM' to our senior team, their eyes just glaze over," he says. "We just talk about how the products need to be managed."
With funding tighter because of the recession, not only does IT need to clearly communicate the business reasons for providing high-quality data through MDM, but the reasons also have to be translated into a specific business cost or opportunity. For instance, instead of stating, "We need a 360-degree view of the customer," a better argument would be, "We need to increase cross-selling by 20%."
That sort of business language "positions things correctly and puts a laser focus on what you're trying to accomplish," Swanton says.
Making the Business Case
One business case for MDM would be to argue that it can help prevent inaccurate order deliveries. If 5% of your shipments are returned because the ship-to addresses were inaccurate, what does that cost in terms of time and labor required to readdress the packages, money for postage and possibly lost customers? "It takes time [to explain a scenario like that], but that's what the business guy will understand," Swanton says.
A second business case is that MDM can help curb procurement costs. Say a manufacturing company uses the same part in 30 plants, but various systems call that part by different names. If data were harmonized, the company could get volume discounts and it would have greater negotiating power.
A third business case for MDM is that it helps avoid data errors. Say you have a new ERP application that isn't working correctly, resulting in failed product runs and an inability to keep items in stock. "Half the time, it's due to [data] errors" that could be fixed via MDM, Swanton says.
There are some telltale signs that data is being managed poorly and you need to undertake an MDM project, he says, and they often have an emotional element. "Usually there's an undercurrent of rage of 'I can't get the reports I want,' 'The system messed up this order' or 'There's been an invoice returned over here,' " he says. "We need to get out of the emotional area, profile the data, look at the complaints and trace the root cause to see what the problem is."
The upshot: "There is no such thing as an MDM project -- there are only business projects where MDM is part of the solution," Swanton says.
Marcelo De Santis, director of enterprise master data at Kraft Foods Inc., can attest to the need to focus on solving business problems. He says that an ambitious MDM initiative that his company began in 2005 lost traction after a couple of years, only to be put back on track when Kraft embarked on a global SAP implementation.
"It changed the perception of MDM because there was an understanding of how important good data was to the SAP implementation," De Santis says. "Before, it was, 'Let's boil the ocean.' Now, it's, 'Let's focus on a specific business need and what we need to do to make that happen and have a way to measure against that.' "
De Santis moved from hanging the business case around MDM itself to discussing the importance of analyzing spending data at all global business units to identify potential savings. He also stressed the importance of quickly absorbing recent acquisition Cadbury-Schweppes. Both projects required MDM under the covers.
Case study: Avon Products Inc.
In 2001, Avon Products Inc. embarked on an MDM initiative to improve management of its product portfolio by creating a common categorization scheme for all of its lines, from beauty to fashion and home goods.
For instance, an offering like scented shower gel might have been categorized as a personal care product in some markets but as a fragrance elsewhere. Now it's categorized the same way globally.
"We went from roughly 10 product categorization schemes to one," says Peter Winters, vice president of enterprise information management.
At the same time, Avon embarked on several business transformation programs, including a move from country-based marketing to one based on clusters of countries, and from local supply chain operations to regional and global ones. Both initiatives required standardizing the free-form product and supplier data generated by local offices, Winters says.
Using MDM technology from Data Foundations Inc., Avon converted hundreds of thousands of items in marketing and supply chain systems worldwide into a standard format.
In addition, Avon "looked at opportunities where we could design major new systems to work with the centralized MDM strategy from the start" as legacy systems were phased out, Winters says.
Now, he says, Avon can analyze product performance, confident that the aggregate data for products, locations and suppliers are standardized and accurate.
A sign of success came after the completion of an MDM initiative in Europe, when it was the chief financial and procurement officers -- not De Santis -- who insisted on driving the effort further into the global organization.
So far, De Santis says, Kraft has made a "significant investment" in MDM, mainly using SAP-based technologies. The effort included improving automation of data management and workflows, establishing data ownership and data quality, and cleansing data from as many as 20 sources to import it into the new ERP system.
Benefits include shortening the process for new product introductions -- because there's a standardized process and a single place to manipulate new data -- and reducing the cost of managing vendor data by having a central place to enter vendor details.
Avoiding the 'Big Bang'
Kraft isn't the only company that learned a lesson because it tried to take on too much too soon. Beniwal says that Daymon began its MDM initiative with the overly ambitious goal of fixing four types of data at the same time: supplier, product, customer and employee.
The company quickly scaled back and prioritized its MDM implementation by focusing on the highest-payback domains -- product and supplier -- with the goal of lowering costs. It took eight months to implement those domains, and the system went live late last year. This year, the company has focused on the customer and employee domains.
Other pare-backs included shelving the idea of providing real-time data to operational systems and instead going with batch updates. Daymon also stuck with English-only implementations rather than local-language systems.
"People want to go for the big bang, but that reduces the chances of success and introduces more risk," Beniwal says. Swanton agrees. "Think a couple of years ahead," he says, "but deliver in six-month intervals."
Brandel is a Computerworld contributing writer. Contact her at firstname.lastname@example.org.
This story, "Untangling your unruly data" was originally published by Computerworld.