Measuring What Matters
Posted by 
Share this Post
Are you measuring what actually matters for data quality in your org?
In a recent post, Jim Harris of OCDQBlog quoted Thomas Redman as saying “It is a waste of effort to improve the quality of data no one ever uses.” Jim also offers something equally as thought-provoking from Thomas Ravn: “There is no point in monitoring data quality if no one within the business feels responsible for it.”
I'd like to heartily agree with these statements, and offer some thoughts on how to address these issues.
Issue #1 is that, out of all the data you manage in your Enterprise Information Management ecosystem, how do you find those data which are actually being used? This is similar to questions I've raised in the past when writing on next generation Data Warehouse architecture (see Eliminating Data Warehouse Pressures with Master Data Services and SOA in last summer’s TDWI Business Intelligence Journal).
In this article, one of the questions I raised was about how many of the hundreds (or thousands) of data elements you bring over into your DW are actually used. The premise is that our DW ecosystems are pushing around significant amounts of data that are never even used.
If you follow that argument, then there must be a ranking of which data are used most often, and those that are rarely, if ever, used. But we are implementing data integration and data quality management for significant amounts of data that don't really matter, and are not being used.
My question is this: out of all the elements that are used, which are the most important? Is it feasible to invest in data quality management for those pieces of data that are the most often used? This overlaps with the analyses we do for MDM projects (My colleague Larry Dubov covers this extensively in his Building a Business Case for MDM series).
This analysis ferrets out the most pressing business problems and the underlying root causes of those problems in terms of master data quality. What would the overlap be between those "most used" data in a Business Intelligence ecosystem, and those master data that are most critical to business success? I would venture that at the subject area level, there would be almost perfect agreement.
Issue #2: Who feels responsible for data? If we could narrow down the list of the most critical sets of data in terms of breadth of use and contribution to meeting business objectives, I'd bet we could find owners for those data who care about the quality of the data.
Part of Master Data Governance is the assignment (election? coercion? enlistment?) of data owners. Those are usually business leaders who have the most invested (the most to gain or lose) in the quality of a set of data. In other words, they have "skin in the game" when it comes to data quality because it positively or negatively affects their ability to meet business goals and objectives.
Issue #3: How do we articulate what those data quality needs and improvements are? Well, if we successfully deal with the first two issues we should have the ability to focus on a discrete set of data elements for quality improvements.
The declarations I've discussed before are the ways we articulate the Whys, Whats, Hows, Whens, Wheres and Whys of data quality. The establishment of Principles tell us why certain data matter to the organization, and Policies tell us what data quality needs are addressed.
If we've done this work, we will have addressed the issues posited in Jim's blog posts. This is where the owners of data get consensus on policies, processes and business rules that must be enacted to meet certain data quality goals.
I believe that data governance is the only place to address these problems, because it's the only time that business leaders from across an ecosystem meet and agree on priorities and make declarations of policy to solve data quality issues and reach corporate business goals
In my next blog entry, I want to address issue #4 - Measuring what Really Matters, where I'll tie in some additional thoughts.
3 Responses »
Trackbacks
Leave a Response
Entries(RSS)
Great post Marty,
Thanks for the blog mentions, but first I wanted to point out that the quote “there is no point in monitoring data quality if no one within the business feels responsible for it,” was actually from Thomas Ravn.
I definitely agree with both Thomases (Redman and Ravn) as well as with you.
Understanding how data is being used is just as important as understanding the business processes that create it and the technical processes that manage it, and without data quality metrics that meaningfully measure what matters in tangible business relevant terminology, no one should either expect anyone within their organization to feel responsible for data quality, or expect anyone to view data as a corporate asset.
Best Regards,
Jim
Jim,
Thanks for the correction. I've updated the post above.
Crysta