In-Depth
MDM: Background, Benefits, and Best Practices
Highlights from a keynote address at TDWI’s MDM Insight conference
At TDWI’s MDM Insight conference held earlier this month, Jill Dyché, partner of Baseline Consulting and conference co-chair, gave a keynote address that drew keen interest from attendees. Since the conference was invitation-only -- limited to executives responsible for planning or approving MDM projects -- we wanted to bring her message to a wider audience.
BI This Week caught up with Jill and asked her to recap the major points of her presentation and shed light on what MDM is, precisely, what benefits are promised, and how best to get started with MDM.
BI This Week: For our readers who may be new to the topic, what is master data and how is it different from the "ordinary data" in my data warehouse?
Jill Dyché: Master data is the reference detail for a specific business subject area. It's common across different business processes, systems, and organizations, but in most companies it's not reconciled between systems. A simple example is that I can go into a bank branch and make a deposit the same day as I withdraw money from an ATM at another branch and pay a bill through my bank's Web site. All three transactions may have separate variations of my "master" data: the non-transactional data that defines a business entity. In this case, that business entity is me, the customer.
What is MDM -- is it a technology, a piece of software, an approach, a philosophy?
Good catch, Jim. It's definitely more than just technology. In our book (Customer Data Integration: Reaching a Single Version of the Truth -- Wiley, 2006, with Evan Levy), we define MDM as "the set of disciplines and methods to ensure the currency, meaning, quality, and deployment of a company's reference data within and across subject areas."
In other words, it's about making your systems "play nice" with one another through their data. So, while MDM is most closely associated with the emerging crop of "hub" technologies that are solving a new crop of problems, it really can't be successful unless, as you say, there's a new approach in place for giving people access to standardized, reconciled, and integrated information.
What kind of information or insight can MDM provide that enterprises don't get now? What kinds of questions can it answer for me as a BI practitioner?
MDM as a business capability doesn't really answer questions -- that's what BI is for. MDM just makes the answers better. By harmonizing common data across systems, an MDM hub can cleanse, enrich, and integrate data that can support reporting and advanced analytics. As a BI practitioner, you can leverage MDM to make your ETL jobs easier and more real-time, and make the data on your data warehouse a lot less latent and a lot more accurate!
So MDM is data quality?
Among other things. Most MDM products on the market offer functions such as value standardization and data correction, which are also common to data quality tools. This type of functionality is "baked in" to MDM solutions. This means that resulting master data is accurate before it's accessible to the systems that need to access it.
MDM goes beyond just data cleansing. There are other functions in the MDM "stack," including application interfacing, grouping and hierarchy support, change logging, and workflow automation. The goal of MDM is to generate and maintain the trusted, "master" version of a record -- be it a customer, a product, or a location. Data quality is core to MDM functionality, but it's just a piece of it.
What other benefits can MDM provide?
Regardless of all the functional and processing stuff I just laid on you, MDM really is a business solution. It delivers authoritative information to common business processes. As such, it can solve a range of business problems.
For instance, we have a media client right now using MDM to support a very aggressive merger and acquisition strategy. It used to take the company an average of 14 months to "onboard" an acquired company's data. Now it takes a week, and most of that is still human time. That's one example of many.
When MDM first came on the scene, we were all focused on the architectural differences, what the best type of hub was, and whether the data should be persisted or not. These days, conversations about how MDM products do the job aren't as amplified as discussions about the business problems they solve.
What gets in the way of adopting MDM?
The same, well-worn cultural and political stuff that sabotages so much change in companies today. People have truly realized that knowledge is indeed power, so they don't want to give up their data or collaborate in its re-definition or consolidation. They'd rather maintain control of their dirty, poorly-defined, incomplete customer records!
This is why executive sponsorship is so important for MDM. Most MDM teams need to enlist someone with the organizational authority to "convince" people [Jill smiles] to participate in exposing their data to the company-at-large, for the benefit of the greater good.
To the same end, we see some IT organizations circling the wagons a bit. It's hard when you've built a data integration capability from scratch, or when you've planned to extend a system to do more, only to be told there are now purpose-built solutions that solve the problem better. We still see people at our clients clinging to their outdated, homegrown solutions that have become far too skills-intensive and expensive to maintain.
There are folks who believe that faster data warehouses can solve the MDM problem. If that were true, we wouldn't have seen all those data warehouse technology vendors acquiring MDM start-ups!
What are the best practices you recommend for an enterprise that wants to get started with MDM?
Well, I'd cite some of what we heard from our esteemed speakers at the MDM Insight event. Barry Briggs from Microsoft emphasized discrete program planning and technology acquisition for MDM. Cathy Burrows from Royal Bank of Canada talked about MDM evolving with the business, first with banking, then with the insurance business, and so forth.
Brian Rensing from Proctor & Gamble had a great quote: "You have to start MDM by attacking a problem that hurts people in their day-to-day jobs -- and use easy words." I loved that last part, because it says everything about making an effective pitch for MDM.
In your Monday keynote at the conference you had your own advice for launching an MDM program. What did you tell people?
I told people to build a road map, we call it an MDM Masterplan, that ties implementation activities to business benefits and shows a timeline for results. This not only helps make the case for MDM, but illustrates how it's going to evolve over time. This is a great way to not only plan the implementation but get the business on board. I'd go so far as to say, don't formally launch any MDM initiative without doing this first.
You're an acknowledged MDM thought leader, Jill, but did you learn anything new at the MDM Insight event that you didn't know before you went?
Well, you know, I was the conference co-chair, so I learned how complicated it was to put together a great conference agenda! [laughs] Beyond that, I'd say I learned the extent to which people are really getting serious about MDM.
The attendees at MDM Insight weren't there to learn the definition of MDM. Everyone had already done their research. They were there to know how to launch an effective MDM program -- one that sustains them for the long-term. So MDM isn't just business-as-usual, and it's definitely not just a flash in the technology pan!
- - -
Jill Dyché will be joined by Wayne Eckerson, TDWI’s director of research and Jill’s co-chair of the MDM Insight event, for a retrospective Web cast on April 15, 2008. They will discuss MDM Insight highlights and present ten lessons learned from the conference. Click here for more information or to register/.