Many Canadian senior-level enterprise architects and data management types turned out for the MDM Canada Summit last week, which featured a panel discussion dubbed "Managing Master Data in the Very Large Enterprise" that yielded the best practices that have come out of the Hudson's Bay Company's and the Bank of Montreal's own MDM experience.
Getting the funding to even get an MDM implementation off the ground can be a challenge, but a good business case is the best place to start, according to Jacob Kuijpers, a senior manager in business intelligence and CRM for the Hudson's Bay Company.
"It's really something that you cook up in the architecture," he said. "When the business side looks at the plumbing, that it isn't there is a mystery to them--they've already spent millions, so why isn't it in there already? This means that we have to think big, and have MDM be the first step. After that, you'll actually have the data, instead of it just being in there."
From there, said Kuijpers, you have to reach out to the various touch points of the project.
"Once you show that having that data at their fingertips is indisposable, then you can build your business case," he said. "It's what you do with the data--you can bring a horse to water, but you can't make it drink. It has to be an access point for all the other projects that we do."
When it came to Hudson's Bay's own upcoming projects, the usefulness of their MDM initiative could be filtered through the necessity for the customer-data-rich applications like an updated gift registry and a new Web site.
For conference attendee Yaj Bhattacharga, an enterprise architect with the nationwide telecommunications firm Allstream, doing things twice seemed silly as well. "We wouldn't want to develop a completely different app for (online) self-service (that doesn't capitalize on the MDM data), right?"
These types of integration issues brought Janet Zilstra, a data analyst with the insurance company Manulife, to the conference. Said Zilstra: "Integration is always an important aspect of any useful system, but it's always painful."
The utility of such a rich data repository, however, could be a good selling point for the executives, according to Henry Ng, a development manager with the Bank of Montreal. Said Ng: "The data coming in is an asset, and MDM is an opportunity to fix (redundancy) and to take a look at ROI. But here, it's ROA: return on asset."
This type of strategy could put a company on the cutting edge: according to a recent survey done by the MDM Institute of over 150 C-level enterprise executives, 60 per cent didn't measure their data governance against processes against their key performance indicators.
"We have to evolve shareholders from data ownership to data stewardship," said the Burlingame, Calif.-based MDM Institute chief research officer and conference chairman Aaron Zornes.
Ng said that it's also important to relate the need for MDM back to past problems in the enterprise so that it has some common ground with unmet needs. An MDM initiative, for example, could be used to reduce the number of data cleansing staff, according to Zornes, or to improve marketing practices via the creation of up-sell and cross-sell lists.
This type of bundling is a common in the telecommunications space, where the battle for customers is getting fiercer all the time. Bhattacharga said that MDM is becoming a crucial application in that space due to its data capabilities.
"We're dealing with matching the correct products and customers, across various market segments," he said. "And we're trying to offer converged services...but it can be difficult to achieve synergy across the different market segments," he said. "We need a common view of the customer--product view is not enough."
The huge amounts of customer and other data stored in an MDM system can give rise to a host of security and privacy issues. Ng suggested calling upon a security officer to establish and reiterate security policies around MDM to those involved with the project.
"You have to educate your stakeholders and staff, and teach workshops around this, especially as new staff might not be familiar with these issues. Have the policies published and make sure that they are updated to ensure that your information is protected," he said.
And there are things to keep an eye out for, too. Said Kuijpers: "As long as vendors keep building into their products that little button that says 'Export to Excel,' security is a misnomer, especially in business intelligence. People just take queries and dump them into their spreadsheets and throw them up on a LAN. (Even with policies) people still circumvent them."
One way to help screen out those without the proper data respect is to utilize data profiles, said Ng. "People might be experiencing the customer data in other ways than you'd expect, so you could try having them need the latest module to see the metadata. This way, the user access matrix could grade your users," he said.
"Plus, a data-scrambling concept can be used if you don't want it falling into the wrong hands. The data doesn't need to be in the hands of the developers all the time. It can be partially scrambled in test."
Related content:
Scotiabank aiming to master data management
Five tips for mastering data governance
Information will exceed storage capacity, says IDC
icon.
