Ten Lessons from MDM Early Adopters
We recap the 10 major lessons learned during TDWI's Master Data Management Insight 2008 Conference
By Wayne Eckerson and Jill Dyché
The Master Data Management Insight 2008 conference, held March 2 - 4 in Savannah, GA, received favorable reviews from the 100 senior-level attendees. MDM Insight 2008 was the first invitation-only event in the MDM industry, and the first topic-specific event hosted by TDWI. The attendees were enthusiastic about the content, which was primarily delivered by practitioners who are early adopters of MDM technology. The event sponsors were MDM thought leaders whose solutions have helped many early adopters deliver reconciled master data to a variety of business applications and organizations. MDM Insight 2008 featured an intimate atmosphere that fostered a rich dialogue among participants and highlighted numerous valuable lessons.
Here is a recap of the 10 major lessons learned during our two-day immersion in MDM in Savannah.
1. Getting Started. Irrespective of their industry, market segment, or existing IT environment, all of the attendees at the MDM Insight event were interested in the right way to launch an MDM initiative. Most had done the due diligence to determine that MDM was the right answer to their business problems. Many had secured budget. Attendees grilled speakers and vendors alike on the best methods for launching a sustainable MDM program. Jill Dyché, partner at Baseline Consulting, says the lesson was clear: "Start small -- with an initial project -- but think large-scale, long-term, and across subject areas." As the early adopters reminded their audience, MDM requires new technologies, specialized skills, and a business focus. With all of these ingredients in place, they said the payoff is well worth the effort.
2. ROI. Perhaps the biggest issue on the minds of attendees was how to justify and get funding for an MDM project. As with data warehousing projects, some organizations are blessed with enlightened executives who understand the correlation between high-quality, consistent data and their strategic objectives; these executives may approve MDM projects without delay. Other project leaders must closely align MDM projects with business need and pain and perform a detailed cost-justification analysis.
Most speakers said that MDM project managers can easily identify and monetize cost savings, but the best approach is to align the MDM initiative with strategic objectives or shoehorn it into approved projects as a necessary underpinning for success.
3. Serendipity. MDM is a business solution that offers a myriad of unexpected benefits once implemented. Many MDM early adopters discovered that once they cleansed and reconciled data through an MDM initiative, they not only improved the quality and consistency of the data available among systems supporting key business processes, but they also enabled other business initiatives such as mergers and acquisitions support, customer relationship management, target marketing, and supply chain optimization.
"Data is a corporate asset, and when carefully managed, it provides a strong, stable foundation to support any information-centric business initiative an organization wishes to pursue now or in the future," said Wayne Eckerson, director of research at TDWI. "Without MDM, many organizations will spend millions of additional dollars executing information-centric strategic initiatives -- or won't even attempt them at all."
4. Change Management. Change management is key. From a technical perspective, understanding and socializing the impact of MDM development activities has everything to do with the perception of success. From a cultural perspective, managing the expectations of business and IT stakeholders is nothing less than a make-or-break proposition. "Change management is hard and can derail an MDM project if you aren't careful," says Barry Briggs, chief technology officer for Microsoft IT. "When you change the data that end users have become accustomed to receiving through reports or other means, it can cause significant angst. You have to anticipate this, implement a transition plan, and prepare the users."
5. Roadblocks. IT and business can pose significant roadblocks. One speaker after another spoke of the challenges of MDM, and they all agreed that IT stakeholders, many of whom are beholden to established technologies, often need to hear the MDM pitch as much as business users do. Distinguishing MDM from incumbent technologies is an often-underestimated step. Conversely, the business may not want to initiate or fund an MDM project when they already have many of the existing tools and technologies required to do MDM.
One attendee asked during a panel discussion, "Our business has already funded our data warehouse and a customer relationship management solution. They are asking, 'Wasn't the data warehouse supposed to solve these issues? Wasn't the CRM system supposed to reconcile customers?' How can I now convince them to take on an MDM initiative?" The panel replied that MDM can optimize those existing solutions, reducing their overall expense and minimizing the risk that they will deliver bad data (which could be their kiss of death). They also said that when purchasing an MDM solution, you shouldn't pay vendors for comparable technologies you already have in house, but which come bundled in their packages.
6. Enterprise Scope. "Enterprise" MDM may be fraught with problems. While the vision of "start small, think big" was pervasive among the event's attendees, the vast majority admitted to wanting to quickly broaden their initial implementation to encompass additional domains, systems, and data. "We started by supporting our CRM program with better data," said Scott Sullivan, CIO and CFO of Pitt Ohio Express. "This let us perform business functions we could never have done with native CRM." Sullivan went on to explain that the company now plans to extend its MDM capabilities to additional operational systems.
Evan Levy, a partner at Baseline Consulting and its director of technology and integration consulting, said that once an organization has implemented an MDM project and it takes root, the organization can then decide whether to "widen the road" by adding more domains to the existing environment or "extend the road" by using MDM to address other business problems.
7. Data Governance. The theme of data governance appeared throughout the conference, showing up in Forrester analyst Rob Karel's presentation -- "Data governance is a critical path to MDM," said Karel -- and mentioned frequently as a common challenge by many of the presenters. Monday keynoter Jill Dyché cautioned the general audience against prematurely launching data governance by forming a committee.
"In order to be effective," Dyche noted, "data governance must be designed. A company's cultural norms, established development processes, and incumbent steering committees must all factor into its data governance framework." Levy recommended growing data governance organically and in lockstep with an MDM architecture, which evolves over time. "First, define what policies and rules you need the business to formulate to support an MDM project that solves a business problem. Then, you can formalize the requirements needed to sustain the initiative. That way, the business is working in their own perceived interest, not IT's."
8. Cross-System Data Analysis. One speaker after another highlighted the time and costs involved in understanding and modeling source data that spans multiple, heterogeneous systems. Microsoft had 10 people working for 100 days to analyze source systems targeted for MDM integration, while the European Patent Office has 60 people analyzing and managing patent data originating around the world. Todd Goldman of Exeros estimated that the services-to-software ratio in MDM deployments is 10 to 1, with cross-source data analysis consuming the lion's share of the services. Just as in the data warehousing world, when early adopters in the 1990s underestimated the quality and condition of source data required to deliver an enterprise data warehouse, many attendees at the MDM Insight event had not thought much about the challenges of understanding and reconciling source data.
9. Matching. Calibrating matches between records generated by disparate systems is both art and science. Many speakers acknowledged that these matching engines -- which are typically delivered within a data quality tool -- are the brains of the MDM system. Barry Briggs of Microsoft said he has three PhDs on his staff to refine the matching and deduplication logic, among other things.
Other MDM Insight speakers said that they'd had a few go-arounds configuring their matching rules. Many shared the consequences of "over-matching" records and thus consolidating two products or customers into one faulty record. The point was that matching needs to be refined over time. In his Tuesday keynote, Evan Levy remarked that "MDM is as much about business and data rules as it is about the data itself."
10. Logical Data Models. An existing logical data model can propel you forward. Conference speakers emphasized the importance of data administration skills in MDM. While some vendors maintain that a logical data model isn't required to make their tools work, most agree that the exercise itself can help a company understand definitions and usage scenarios for enterprise data, which makes it easier to gain consensus around data policies.
Carl Gerber, senior manager of data warehousing at Tween Brands, shared how his company's data modeling and stewardship skills were a large part of his team's successful MDM delivery. Tween has created a series of data integration hubs to manage various subject areas, such as product, inventory, suppliers, and merchandising hierarchies. All data exchanged between systems passes through these hubs to ensure data consistency and reconciliation. The architecture has eliminated numerous point-to-point interfaces and improved operational efficiency, decision making, and revenue-generation objectives.
- - -
Wayne Eckerson is the director of TDWI Research. Jill Dyché is partner of Baseline Consulting (http://baselineconsulting.com/). Wayne and Jill were co-chairs of the MDM Insight 2008 conference.