And that's not a good way to start an MDM implementation, no matter whether MDM is part of a data migration for a new software implementation or part of a consolidation of ERP (or other enterprise) applications. These are the two most common types of MDM initiatives, according to Bill Swanton, vice president of industrial products research for Boston-based AMR Research.
Costs for MDM initiatives may rise when an MDM project is not scoped properly at the outset. Without proper scoping, a manufacturing company that budgets MDM as part of its ERP implementation may think that 5% of the total project cost will be sufficient for its MDM initiative.
The company often ends up spending 15%, however. "If they find out too late in the budget cycle, that means a 10% overrun and angry users who find data problems after go-live," Swanton said.
There's a relatively easy way to avoid this. "As part of a data migration, you need to build a repeatable process for extracting data from the legacy systems and finding quality problems that will affect the operation of your new ERP or other system," Swanton said. "Then you need to create a business process that ensures that new master data as well as changes to master data are valid after go-live."
An MDM initiative may also be needed when an organization's application consolidation effort does not include data harmonization – an omission that can lead to trouble tickets and complaints about data quality. When that happens, "slowly but surely, processes are put into place to hold the business accountable for data quality," Swanton said.
Even if a company manages to get top-level CEO mandates on board for an enterprise-wide heterogeneous MDM initiative, it's hard to take a "big bang" approach to MDM, for three reasons: cost, managing new personnel-related processes, and the onerous task of physically manipulating vast amounts of data.
But the slow, incremental approach to MDM implementation can also create problems. If a company moves too slowly, some of the earlier projects or pieces of the overall implementation – servers, applications, software versions or even business processes -- could turn into legacy solutions that would require support even as the enterprise-wide MDM solution evolved. One solution is to frame the incremental approach to MDM within a long-range plan.
"Without a strategic plan for designing a comprehensive master model, a project-by-project rollout may introduce spurious dependencies that may be difficult to ignore down the road," said David Loshin, president of Knowledge Integrity, a Silver Springs, Md.-based information management consultancy.
Loshin recommends instituting the proper data governance practices, including a structured process for analysis of data requirements, as well as managing data quality. The process should supersede and augment the traditional siloed, project-based system development lifecycle. "This should clarify where the benefits of MDM lie," he said, "and it will help in scoping out the MDM program."
Create a map of enterprise data
A company embarking on an MDM project should start by creating a map of its enterprise data. After all, a company has to understand where its data is and how that data relates to other data, applications and business processes.
"Whether a company uses a data discovery and mapping tool or begins by surveying users to ask them 'manually' about their information access and use patterns and behaviors, knowledge about current data is the first step towards truly effective MDM," said Michael Dortch, principal analyst of DortchOnIT.com.
"You can't manage what you can't measure, and many enterprises have few or no resources devoted to measuring or monitoring how users use the information that enables and drives the business," Dortch said. "It doesn't have to be as elaborate as a true, interconnected real-time data map, but the closer an enterprise can get to that goal as a start, the firmer the foundation of that enterprise's MDM strategy will be."
Starting small, it turns out, doesn't necessarily mean going slow. Moving too slowly can result in data dependencies that end up hindering future progress.
"Most manufacturers start small," said Andrew White, a vice president of supply chain management research for Stamford, Conn.-based Gartner Inc., "perhaps with one domain -- the product. Then they'll sequence the second domain -- the customer -- the following year."
White recommends that manufacturers also start with operational MDM, where data is created initially, and then move on to analytical MDM for business intelligence (BI) purposes.
When it comes to scoping an MDM effort, White recommends that users cover all the bases of an IT project:
- Vision -- a big picture explanation of how MDM supports the business strategy
- Strategy -- how that vision will be realized over time
- Governance -- decision rules, rights, regulations, responsibilities
- Process -- the workflow of who does what and when they do it
- Organization -- the who and where of reporting lines and responsibilities
- Metrics -- measuring success of data, processes and the entire MDM program
- Technology -- what it is and what it's expected to achieve
Wanted: MDM and PIM consultants
But to realize this vision, many manufacturing-based organizations will need to hire outside consultants with MDM and PIM experience. "It's much like an enterprise CRM project or enterprise data warehouse," said Aaron Zornes, founder and chief research officer of the MDM Institute. "You're probably going to need some outside assistance, possibly to do the evaluation, and [probably] the implementation as well."
MDM is especially complex at the enterprise-wide level. "If you're larger, it's a bit harder because you've got one solution from company X, one from company Y and another from company Z," Zornes said. "And you'll probably need an independent hub," he said. This hub would be a third-party master data hub that correlates data in a central database.
For smaller companies embarking on an MDM initiative, Zornes has good news. "A number of ERP and CRM [vendors] have started bundling MDM/PIM into their solutions. It's often integrated -- you just have to turn it on. Ultimately, if you're a midmarket user, you will get MDM bundled in."
Master data champions
Even if the possibility of acquiring MDM built-in exists, there's little point in starting an MDM project without executive support and a master data champion. If you don't already have executive support, one way to get it is to identify the "low-hanging fruit" that could be plucked by an MDM rollout.
Through global spend analysis, manufacturers can find out how much they are spending with a given supplier across all of their organizations, Zornes said. A manufacturer might have several plants that operate independently with the same core suppliers. By recognizing items that are purchased across an entire organization, a business can renegotiate with suppliers to achieve immediate cost reductions for higher volume orders.
"This could be a BI project, too," he said. "But, in practice, you need it integrated with MDM to keep track of suppliers as well as services."
"It's hard to believe, but companies with multiple divisions still don't talk to each other," Zornes added. They don't know which divisions share customers or which divisions have complementary products. For manufacturers that have grown through mergers and acquisitions, this can be a big problem -- but one rich in opportunities that can be revealed through MDM projects. "They purport to be one company, but they don't work together," he said.
Chris Maxcer is a freelancer writer who frequently contributes articles to SearchManufacturingERP.com.