By using website you agree to our use of cookies as described in our cookie policy. Learn More

TDWI Upside - Where Data Means Business

The Future of Data Governance

What might a governance model for data science and exploratory analytics look like? To get an idea, head down to your local Trader Joe's.

What might a governance model for data science and exploratory analytics look like?

To get an idea, head down to Trader Joe's. With its curated selection of private-label goods, carefully arrayed and displayed for shoppers to browse -- in some cases, to sample -- Trader Joe's can be a model for a new, consumer-oriented governance regimen.

For Further Reading:

Adapting Data Governance to Big Data

Data Governance Doesn't Need to Be Gatekeeping

Governed Data Discovery: Blending Self-Service and Standards

"The ethos of Trader Joe's was to target Ph.D.s who were fresh out of graduate school. These were people who were well-educated but who at that point in their lives didn't have a lot of money," notes Donald Farmer, founder of TreeHive Strategy, an information management advisory firm. "That probably doesn't describe most data scientists, however."

Rethinking Governance

At TDWI's upcoming Chicago conference, Farmer will teach a class ("Governance in the Age of Self-Service") in which he'll outline a new, consumer-centric approach to governance.

He calls this "data shopkeeper" governance.

Farmer sees Trader Joe's as an excellent model for what he means by data shopkeepers. "If you go to Trader Joe's, you can't buy just any brand of spaghetti sauce. You buy Trader Joe's private-label sauce, but it's really good! That's what the data shopkeeper is doing. They're giving you a preselected feed. There won't be other feeds you can choose from. Their feed will have what you need, but you're going to have to customize it yourself."

Data shopkeepers prioritize access, not restriction. They also emphasize truly excellent customer service. Today, for example, if a company's marketing analysts need customer data, they have to either get the data they need from the warehouse -- assuming it's even there -- or (more likely) compile their own working data sets. In the data shopkeeper model, whoever owns/manages the requisite data -- IT, shadow IT, the line of business, etc. -- would prepare the data feed for the customer that requests it.

The data in this feed would be cleansed and quality controlled. The analysts would all have access to the same feed and could use it to create their own custom data sets.

Think of this as analogous to Trader Joe's private-label spaghetti sauce: it's preselected and quality-assured; it has most of what you want in a sauce, but you can always tweak it.

"You get your data set and it's like the IT department's private label," says Farmer.

Legacy Governance Has to Go

"The status quo is that you have [a legacy governance model] that's obsessed with restricting access and centralizing control. The data shopkeeper [model] is still regulated and still enforces compliance, it's just much more about serving the needs of customers," he says.

Farmer and other experts contrast this new, consumer-oriented governance model with its predecessor -- the so-called gatekeeper model. "If you look at how business intelligence evolved over time, it evolved out of an IT function, with IT as the gatekeeper. BI has always been centralized, usually under IT, although recently it's moved out of IT and into a center of excellence. In this case, too, there's still some degree of central control around it," says Chris Adamson, president of information management consultancy Oakton Software.

Adamson, a veteran TDWI instructor, addresses this issue in several of his courses, including a session ("Data Modeling in the Age of Big Data") he's slated to teach at TDWI's upcoming Chicago event.

"This centralized model fits with that notion that there's a gatekeeper -- usually, an IT gatekeeper -- whose priorities aren't the same as [those of] the business. The gatekeeper's priorities are repeatable processes and predictable change management," he says. Of course, repeatable processes and change management aren't quite as important to data scientists. This is why we're mistaken, Adamson argues, if we think we can govern data science and similar exploratory practices like we did the data warehouse. "That would be a disaster with analytics," he insists, arguing that "we cannot attempt to impose the traditional process on developing analytics. We'll ruin it if we try and treat it like everything else."

Consumer-Oriented Governance Is Not a Contradiction in Terms

It might not be obvious at first glance, but the new consumer-oriented approach to governance championed by Farmer, Adamson, and others isn't a free-for-all.

Farmer once again uses Trader Joe's to make this point. Not everybody can buy everything at Trader Joe's: shoppers under a certain age can't purchase alcohol, for example; nor will staff serve samples to toddlers or children without first getting permission from their parents. There is regulation and control. There's strict compliance, too -- albeit with a very different set of standards.

"It doesn't mean there's no regulation, that the shopkeeper has no control. It's actually highly regulated. Trader Joe's is a great example: they make their store very welcoming, and they have a very clear idea of who they want to market to and how they want to market to them. They won't sell beer or wine to just anybody," he points out. "It's just a different form of compliance."

TDWI Membership

Accelerate Your Projects,
and Your Career

TDWI Members have access to exclusive research reports, publications, communities and training.

Individual, Student, and Team memberships available.