Blog: Jill Dyche

« Outsource the CFO! | Main | Data Governance as Subversive Act »

EDW versus MDM: Scotiabank's Solid Strategy

In which Jill relates the philosophy of a data warehousing and MDM best practice--MDM newbies take note!--implicitly admonishing smart people not to buy her martinis.

I recently spent some time with Neil Freake, manager of BI development at Scotiabank’s Enterprise Data Warehouse. I’d met Neil last year at the DataFlux User Event in Vegas and we shared a nice conversation about MDM strategy, some stories about data ownership challenges, and a couple martinis. My memory of the details is obviously hazy, so it was nice to catch up with Neil a few weeks ago in Ottawa and in Toronto, where we were fellow speakers at an MDM executive briefing hosted by DataFlux and SAS Canada.

I’ve been bemoaning the fact that we’ve recently watched a few data warehousing teams circle the wagons, protecting incumbent technology platforms and headcounts and claiming that they’re “already doing” MDM. In doing so, they sabotage organizational understanding and delay adoption of MDM capabilities that could actually enhance the data warehouse—not to mention strategic business programs.

Scotiabank is in the all-too-rare and refreshingly visionary position of supporting its MDM efforts through the data warehousing group. Neil has a realistic and practical vision of how the bank’s MDM strategy overlaps, but doesn’t represent, its enterprise data warehouse capabilities.

“Our data warehouse is requirements-driven,” he explained to a packed auditorium at SAS Canada headquarters. “We only pull about 30 percent of our corporate data onto our enterprise data warehouse. We consolidate data marts, integrate enterprise data and provide information in near real-time, but we are not operational…we’re not systems. That’s how we keep our business users happy, and it’s also how we manage growth.”

Neil echoes the view of many EDW best practices—Scotiabank won a TDWI best practices award back in 2002—that the data on the warehouse should be informed by the business’ need for analytics and BI, rather than a general-purpose dumping ground for all the data ETL programmers can get their hands on. “We’re not going to put a bunch of data onto a data warehouse or an ODS that we don’t need for BI,” he says.

Notwithstanding its data evolution, Scotiabank’s needs for MDM (specifically CDI) are quickly emerging as the integration and reconciliation of customers across systems promises to augment initiatives from target marketing to anti-money laundering. Data warehouse professionals are uniquely positioned to see the benefits of operationally-integrated data through MDM, and usually have the incumbent skills to support operational integration and deployment of master data.

“The EDW can help start MDM,” Neil Freake concluded. “But we can’t fix it, and we can’t be it.” Scotiabank’s MDM program—supportive and mindful of, but separate from the data warehouse—is well underway.

Technorati tags: Master Data Management, Scotiabank, customer data integration, CDI

Share: