Last week’s TDWI Master Data Quality and Governance SuperWebinar represented an on-line reprise of the live conference—co-chaired by yours truly and TDWI Sr. Research Manager Philip Russom — this past April in Savannah. Obviously the on-line format doesn’t lend itself to one-on-one sessions or audience Q&As of the live event. So, herewith I’m taking a stab at some of the on-line questions submitted from webcast attendees.

Donna: Our IT group cannot seem to get energy behind ownership of the IT role in governance and IT data and metadata. What can I do to stimulate the IT side of governance?

Jill: Okay, Donna, I sense from your question that the business is pushing data governance at your company and IT is reluctant to jump on board. So I’m going to give you a working answer and a cynical answer. You choose which one’s most apt for where you work.

The working answer is that the business should engage the highest levels of IT — probably the CIO — and explain the problem(s) that launching a data governance effort promises to solve. Remember in my keynote I discussed putting executives in the position of saying “no”? That goes for IT executives, too. Woe to the CIO who breaks the news to beleaguered business leaders that IT can’t deliver.

Here’s the cynical answer: Invite Talend, Initiate (IBM), and Kalido (the esteemed sponsors of our SuperWebinar) to brief you about their data management solutions. Invite IT management to the briefings. Once you’ve been briefed explain to IT that as the business launches data governance it will likely acquire data governance enabling solutions as well. IT gets itchy when the business acquires its own technologies. This might get them to the table faster than you can say, “data as a service!”

Satish: Do you keep the bad history data after you clean up the data in your master hub?

Jill: Several of our clients do indeed hang on to the bad records resulting from their MDM matching. For instance your hub might deem three different customer records to be the same person, and it might earmark a specific record as the best one. Retaining the other records will give you an audit trail and let you refine business rules so you understand what the determining factors are in match selection. It will also let you measure improvements in your data as survivorship issues become clear and fewer anomalies emerge. (I’m assuming you’re talking about keeping a history of non-surviving master records, and not historical data, like you’d put in a data warehouse.)

Vicki: Any tactical advice on how to start implementation of data governance?

Jill: Beyond my webinar admonishments to pinpoint a business problem first? Not really, Vicki. We’re already seeing the fallout from companies that haven’t linked data governance to a business problem, and it ain’t pretty. But let’s assume your challenge is more about data management than data governance, and you need to formalize some practices around data quality, cleansing, validation, standardization, and matching. Let’s say that’s the issue, ‘kay?

I still say pinpoint a business problem first. This is because as you do your discovery, you’ll see how bad things really are. IT doesn’t have a bona-fide data correction process, do they Vicki? People are manually reconciling records all over the place, aren’t they? There’s no metadata (that anyone knows about, anyway) and there are pockets of data modeling going on in different departments. Your company is over-investing in effort, Vicki. Quantify how much what I call these “campfire” work efforts are really costing, then pitch data management as a foundation for data governance.

But still link it to a business problem.

Chris: I’m a Data Architect responsible for creating a Data Management Program. One thing I have been challenged with is developing ROI. Any ideas or experiences you have seen in creating ROI around such a program?

Jill: Chris, did you read my response to Vicki? Quantify some of the costs associated with campfire data management. You won’t be able to quantify them all, but find some juicy ones. Like the fact that your finance department is making over 175,000 adjustments per month before it closes the books. And even then it has to explain to regulators why the books say what they say after they’re closed. And in the dark of night, someone in finance knows the books shouldn’t have been closed, and he’s got night sweats. (Okay, that last part doesn’t have a quantifiable cost, but I think it drove the point home in a dramatic way, don’t you?)

Anyway, once you’ve quantified the human labor costs of not having formal data management, whip that up into a presentation with some hard-hitting calculations and call the CFO. He’ll take the meeting and fund your Data Management Program. He might even buy you lunch—as long as it meets corporate expense guidelines.

Wendy: Who is best charged with building and writing data governance policies?

Jill: You are, Wendy. Because you’re the only person in your group that’s smart enough to have asked that question. So by default, it’s gotta be you.

Seriously, crafting policies on behalf of enterprise data is the job of the data governance council or board. I’m going to defer the policy question to Marty Moseley, who wrote a great blog post on it a while ago. Read it here.

Arun: Hi — I’m a solution architect for DW/BI and would like to know the basic hierarchical structure of a Data Governance Board?

Jill: Hi Arun! Craft an org chart and put yourself at the top and see if anyone notices. This tactic has been effective throughout history. I say give it a whirl.

If you don’t have the intestinal fortitude then I have to tell you that the makeup of your data governance board will likely change over time as new business strategies and priorities emerge and executives become tasked with overseeing them, and the information that enables them. So to answer your question, there’s really no fixed hierarchy. Having said that, the person initially in charge of your data governance council has a lot to do with the council’s perceived credibility. So assuming you yourself demur from donning the mantle [editorial comment: DO IT, Arun! DO IT!] then make sure that the council chair has the organizational authority to make governance stick.

Jill also blogs at JillDyche.com.