Is It Time to Decentralize Data Quality?

Loraine Lawson

I once was in charge of the Kentucky Transportation Cabinet's website. It was in the early days of the Internet, when businesses were becoming aware of the Internet as an e-brochure for the public - and possibly an opportunity for one day conducting actual business.

 

Until I came on board, IT controlled the website. Predictably enough, someone had done something stupid and embarrassing on the site. I can't remember what, but frankly, the animated flashing gifs and multicolored text were embarrassing enough.

 

So, the website was moved into the domain of the Public Relations department, and since I had the right mix of journalism, writing, design and basic HTML experience, I was hired to oversee it. I was also expected to write speeches, help with special events and handle press calls, because, really, how much time could this Web thing take, anyway?

 

You can probably guess how that worked out. Frankly, it was an overwhelming task that I would handle much differently now. But in my defense, nobody had really figured out how to manage a large website that spanned multiple pages and multiple purposes-not even Gartner. I know, because after I (very soon) moved on to a new job as a tech journalist, I wrote a lengthy white paper on the topic and interviewed a wide range of experts. I learned I wasn't alone in struggling to transition websites from the tech department's plaything to a resource for reaching customers.

 

Now, of course, we know much more about it. We'd never think of having only one person responsible for writing, designing and maintaining an entire, multi-layer website. If I could go back, I'd have a much more decentralized management structure, but with a central committee overseeing guidelines about style, structure, standards and e-commerce.


 

Lately, I'm wondering whether data quality might benefit from a website-management history lesson.

 

Data quality is primarily considered an IT function, which makes sense on the surface, given that quality is closely connected to other IT functions, such as integration and storage of data.

 

Certainly, IT seems to think data quality belongs to technologist. A recent survey by Forbes Insights titled "Managing Information in the Enterprise: Perspectives for Business Leaders" found that 79 percent of IT managers say data quality is their responsibility. Interestingly enough, 74 percent of the finance, sales and marketing respondents felt the same way. A recent CIO.com article highlighted this quote from the survey:

While IT managers largely concede that information is the users' not theirs, they take the position that data and information management systems are under IT's purview. This differing perspective puts IT and business executives in conflicting camps, particularly when it comes to data quality.

In reality, IT is doing the work of data quality. The survey found that 51 percent of IT managers say their companies are engaged in data quality-management work, but only 25 percent of business executives said their companies had data quality-management projects.

 

But just because IT has thus far managed data quality doesn't necessarily mean that should always be the case. After all, the first corporate Web pages were done by IT because IT had the tools and understood the technology.


Perhaps data quality, just like websites in the 1990s, is becoming too massive and pervasive for any one department.

 

Joyce Norris-Montanari discussed this possibility recently on DataFlux's Community of Experts blog. Norris-Montanari is president of DBTech Solutions, which specializes in business intelligence and data management. She's also co-author of the book "Data Warehousing for E-Business."

 

She points out that most companies really aren't ready for a huge, enterprise-wide data quality initiative:

Centralized data quality would definitely come from the enterprise data management group, but most of these groups are not thinking about master data or corporate data. Do these people have time to manage a platform for data quality and data integration for the corporation? I think not! So what that means is that most of us are not ready for a corporate initiative, and are really not sure how to go about implementing such a "gorilla."

Norris-Montanari has a quirky way of expressing herself that leaves me wondering where, exactly, she stood on the issue, but I think her key question is telling. She asks whether any of the following suggestions would be helpful:

  • Enterprise guidelines for departmental data management.
  • Enterprise data quality and data integration software, including a single platform for all metadata.
  • People who really understand the concepts of data management for the enterprise.
  • A roadmap that breaks data quality into smaller pieces, for easier implementation.

 

Certainly, those guidelines would've been useful in the 1990s when websites were moving out of IT's sole control and into the domain of the entire business. Is the same true for data quality?

 

I don't know. I certainly think it's worth discussing. There is, however, one thing of which I'm certain: I'm glad I'm not in charge of figuring it out.



Add Comment      Leave a comment on this blog post
Apr 21, 2010 4:59 AM Joyce Norris-Montanari Joyce Norris-Montanari  says:

Loraine - I am responsible for figuring it out for my clients (in my own quirky) way.....  I think small bites will work!

Reply
Apr 22, 2010 10:32 AM Ken O'Connor Ken O'Connor  says:

Hi Loraine,

Interesting article.

I agree with the idea of "divide and conquer". 

I also like the idea of "A roadmap that breaks data quality into smaller pieces, for easier implementation."

Over the course of many data intensive programmes, I encountered the same issues time and time again.   There was a tendency to refer to these as "data quality issues", which of course they were.  However, there are many different types, and even more causes of data quality issues.

I list these on my blog, together with a process for assessing how well your organisation deals with each type of issue.

Rgds Ken

Reply

Post a comment

 

 

 

 


(Maximum characters: 1200). You have 1200 characters left.

 

null
null

 

Subscribe to our Newsletters

Sign up now and get the best business technology insights direct to your inbox.