Teasing out the need for master data management

Candid conversations with executives reveal the sometimes-dire consequences of dirty data

A  few years ago when marketing executives claimed that their data wasn’t ready to support target marketing, everyone nodded their heads sympathetically. It was hard. No one knew for sure where to start. Customer data needed to be cleaned up and reconciled. IT was busy.

Nowadays, those head-nodders are more likely to put their hands on their hips, thrust their heads back, and ask, "Well if not today, then when?" Add executives in finance, sales, operations and strategic planning to those with data integration and governance challenges, and marketing suddenly finds it has plenty of company.

I thought I'd share a few recent conversations I've had with executives who are confronting the need to manage their companies' master data. These conversations were surprisingly candid and refreshingly devoid of excuses. They manifest a cultural impatience with dirty data and an awareness of its sometimes-dire consequences. Any of these ring a bell?

"Our executives have talked about quantifying customer value for a dozen years now. They've read about the other banks doing it, but we know these other banks' secret: They're doing it with duplicate customer records. If I don't know that three customer records are really the same person, then I'm not accurately calculating that customer's true value. Honestly, if we don't apply master data management to this problem, we may as well give up customer value altogether and use that 'strategy of nice' that everyone falls back on."

"I know other CPG [consumer packaged goods] firms have this problem, but having different prices for the same product is embarrassing. And I'm not talking about anything sophisticated like value-based pricing here. I'm talking about having 13 different product catalogs that contain not only different product prices but different size, weight and dimension details. Now some folks in yet another line of business are thinking about building their own product catalog, and I'm trying to convince them not to. I mean, we already have 13 – can't they use one of those?"

"We've stopped using the term 'single version of the truth' around here. Nowadays, that phrase just falls flat. The sad fact is that we have multiple versions of the single version of the truth. I'm not sure what's funnier here, the concept of a 'single version' of something, or the concept of truth. There is definitely no truth. So now let's talk about our ability to have an informed negotiation with one of our suppliers. We can't do it – full stop."

"We're like any other health care provider, except that we're a children's hospital, which means our patients are little kids who have cancer and heart problems and other bad stuff. So with a mission like ours, you'd think that reconciling patient data across service lines would be pretty critical. And everyone insists it is. But no one knows exactly what the answer is, so we're moving slowly. Too slowly. And someday really soon something bad is going to happen because we didn't individualize a patient quickly enough. Am I making my point?"

"We need to change the way we do things internally. We need to go from gut instinct to scientific, rigorous decision making. We don't know why we're putting what we're putting on our websites. Our lines of business accuse us of flawed research, but they can't tell us why. We don't know what customers they have in common. In short, we don't know which ads are working, and we don't know who to market and sell to. Ultimately this is going to damage our brand."

Savvy data practitioners can actually parse these conversations into master data management requirements – if not outright business cases. The trick is to illustrate the path between business challenge and the right set of solutions. And an MDM hub is merely one component.

In our book Customer Data Integration: Reaching a Single Version of the Truth (John Wiley & Sons), Evan Levy and I talk about narrowing executive focus to specific business needs that mandate MDM. We present a series of industry-specific business needs. We also discuss the business' varying needs for data access.

We explain that while data warehouses have long been the de facto solution for integrating data for the purposes of analytics and reporting, MDM tackles a different problem set. MDM is less an end-user-focused solution and more of an active data provisioning mechanism to various operational applications.

Need both? First know what you've got. Document business requirements and build some use cases. Understand how much specialized development you'll need to do, and whether you've got the chops. Present the playbook back to executives. Then talk to some data integration vendors.

The extent to which executives who will be funding data governance and MDM efforts "get" how they propel the organization forward – with all due respect to our incumbent systems – is the extent to which data integration will get traction. Better late than never.

Bio: Jill Dyché is Vice President of Thought Leadership at DataFlux. She is the author of several acclaimed books and the popular blog at www.jilldyche.com. Dyché is the co-founder of Baseline Consulting, which was acquired by DataFlux earlier this year.

Jill Dyché
Vice President of Thought Leadership, DataFlux