Q&A: Keys to Data Governance Success
The surge in interest in data governance points to increased need to manage data better. Why do enterprises still struggle when implementing a governance program, and what can they do to fix the problem?
- By Linda L. Briggs
- January 19, 2011
Implementing a solid data governance program takes a commitment to change management, a solid business case, and ongoing work. In this interview, John Ladley with IMCue Solutions explains why companies continue to struggle to implement data governance, even as the need for it has become painfully obvious to both IT and business leaders. "Now that the light bulb has gone on and everyone is blinded by insights," Ladley says, "everyone wants to do data governance."
Ladley, the author of Making EIM Work for Business -- A Guide to Managing Information as an Asset (2010, Morgan Kaufmann), has spoken and written widely on the topics of data governance and enterprise information management. He has 30 years of experience helping both IT and business leaders to better plan and manage information and projects. Ladley recently spoke at a TDWI Webinar on Data Governance for Business Leaders.
BI This Week: You've spoken and written extensively on the topic of data governance specifically, and enterprise information management in general, for years. Why is there so much interest in data governance right now?
John Ladley: For various reasons, people are realizing that none of the many sub-disciplines within this area I call "enterprise information management" will work at all without data governance. While data governance used to be a nice-to-have element or a success factor, as we learn more about our technologies, and as we learn more about data management, it's readily apparent that it just will not work without data governance. Now that the light bulb has gone on and everyone is blinded by insights, everyone wants to do data governance.
What's your definition of data governance?
To answer that, I'll quote from a slide [in my recent Webinar with TDWI]. "Data governance is the organization and implementation, policies, procedures, structure, roles, and responsibilities which outline and enforce the rules of engagement, decisions, rights, and accountabilities for the effective management of information assets." That's a definition [I prepared along] with Danette McGilvray, Anne Marie Smith, and Gwen Thomas.
Let's talk about a few of the benefits of good data governance -- points you touched on it in your presentation.
Let's look at it from a business case point of view, in terms of pure, hard benefits. When you implement a data governance program, you implement a structure and framework that will put programs into place to lower costs caused by misuse of data or by fraud -- mistakes that are generated by data quality problems. You improve the balance sheet by providing the mechanisms to identify liability and risk in your information.
I think you also improve productivity by a couple of additional, simple factors, although they are harder to measure. First, you have fewer meetings. I've not been able to put a number on it, but I have documented that organizations that implement governance have fewer meetings in which they argue about what to do about data and who does what -- simply because they have principles and policies in place.
Companies are more productive because they're more efficient. In the long run, you shouldn't have inappropriate copies of data everywhere. You should only copy data as it's required for performance reasons or redundancy reasons. You shouldn't have 50,000 [Microsoft] Access databases with the same data in them all. You shouldn't have customer data leaving the organization on a thumb drive.
Those are hard benefits from a pure business standpoint, but some of those are also soft benefits. Productivity is a soft benefit. The sense of security in guarding an important asset is an intangible benefit, but it's certainly important. I think you also have intangible benefits as a publicly traded company. Eventually your market value or your balance sheet goodwill will increase because you have better control of this major investment we call data. If you take steps to eliminate that harm, you eventually have a better sense of goodwill, a better brand, or a better market value.
You mentioned several return-on-investment kinds of reasons for data governance. Are there others?
Mistakes from data quality cost American companies hundreds of millions (maybe billions) of dollars a year. In the last two years, we've worked with three large Fortune 1000 clients. Without much effort at all, within each client, we found seven- and eight-figure benefits for cleaning up data. These were [problems] that they didn't even know were happening.
If you don't have governance and you don't have information management, you most likely have a problem, period. Almost 100 percent of the time, I will find waste in [client] companies. I will find mistakes being made on a daily basis that are costing companies real, hard dollars.
When you work with a company to try and help them put data governance into place, what issue most often derails the approach? Is there a single most common error?
One of the reasons companies don't have governance in the first place is they've skipped over it.
In an attempt to save money?
Yes, or get the project done sooner. That's the superficial reason.
Years ago, we did something we called IRM, or information resource management. … [We would build a data model and so forth], then build a data warehouse. We would then discover that the data coming into the warehouse was somewhat questionable. We would look into why it was questionable, and we would invariably find out that at some point in time, someone had designed a system in which they had identified these exact data issues … as potential problems, and the organization had consciously said, "We'll come back later and fix it." Of course, "later" never happens in a corporation. I put that right up there with, "The check is in the mail."
What we found was that someone had written a data model, which, of course, assumes that you follow the standards within that model. No one had followed the standards, so in the early days of data warehousing, a big question was always, how do we make people follow standards? That, in essence, is a form of governance.
We found that when we showed people standards, they would invariably say, "I don't have time for that," or "That will slow the project down." When you do a root-cause analysis on it, however, you discover that people simply don't want to change the way they do things. That's a factor of human dynamics and human psychology. We are all averse to change.
We've discovered that the No. 1 resistance to data governance is a resistance to changing the way things have been done for 20 or 30 years. It's easy to say, "That's going to slow us down, so we'll get to it later." When you really boil that down and examine cause and effect, however, that's a horrible excuse.
We once had a client where someone said [about a proposed data governance effort], "That's going to slow me down." I said, "Do you mean that if we don't do it, this project will be on time and under budget?" People had to laugh -- this organization had not been on time and under budget in years. They had to laugh because they realized the absurdity of the comment. We're talking about IT here, where 80 percent of all projects are not on time and under budget, and [that track record] has nothing to do with governance. Claiming they will be delayed is an excuse to avoid change.
The fact is that governance will improve things, but it is truly hard for people to embrace a new way of doing things.
That leads to the next question: How important is change management as part of all this?
I think change management is the single most important step [to include in] data governance -- you need a formal, official change management program that sustains the data governance program. It has nothing to do with technology; nothing to do with tools. It has everything to do with organizational effectiveness and organization management.
How do data stewards fit in with data governance?
I use "stewardship" as a very generic label for the entire management structure of data governance.
Data governance is a management program, correct? Any management program, whether it's marketing, accounting, inventory control, procurement, purchasing, or human resources needs someone who's accountable for the success of that area. Data governance is no different. If you set certain standards and policies that need to be executed, you need a person who's accountable -- some party or parties that are responsible for carrying out those policies. You need to identify everyone else who's involved as well, whether they just touch the data or they need to know about it.
I've had some lively debates with clients about this. "Come into our company for a few days and help us identify the data steward," they'll say. If you are a $20 billion consumer-products company, it's not that simple. I'll ask, "Who wants to be accountable?"
"Well, the VP of sales wants to be accountable."
"OK," I'll ask. "Does that mean the VP of sales is willing to be fired because somebody made a big mistake on data?"
Then we get to the core issue. "Well, no."
I have to point out, "Then they don't really want to be accountable."
The first key to stewardship is not so much who is the steward but who is accountable. After that, who bears responsibility for the ongoing quality, and enforcement of policies and procedures? From there, you build out your stewardship organization.
How do master data management (MDM) and data governance work together?
They go hand in hand. You cannot have MDM without data governance. If someone argues with that, my position is: "Why do you do MDM?" The answer is, "We don't know who our customer is," or "We don't have good product numbers." It is common to hear a reason such as, "We have something like 14 customer databases." Why? Usually, it's because no one adhered to data standards, and what's that? That's data governance. You're doing MDM because you didn't have governance, and if you don't have data governance and you do MDM, you will simply have yet another customer database, or another item database, or whatever it is you're trying to address. Data governance is mandatory for MDM.
It sounds like data governance comes first, then MDM, at least in a logical progression.
Well, in an intellectual sense, yes, but in a practical sense, when you implement MDM, you also implement governance, even if you don't call it that. You might call it "data quality oversight" or something like that if the term "governance" is hard to sell, but again, you must have accountability and responsibility defined via policies and procedures for MDM to work.
You'd be surprised how many people try to do MDM without data governance. I have to be careful because I don't want someone reading this to think that I'm talking about them, but we had a client that spent $25 million on an MDM product and an MDM project, and at the end of its painful implementation, did not have any governance nor any business reason for anyone to use it.
Listening to you, I hear a vast general resistance to data governance in the enterprise. Does it mainly go back to a resistance to change?
It all comes back to resistance to change. It's not rational. You get a room full of business people and IT and everyone will agree, "Yes, the data stinks." Then they'll say, "Yes, this is why the data stinks," and further: "Yep, we have to fix it." Then I'll say, "Here's what you need to do to fix it," and they'll say, "Ooh, that's too painful. Let's do that next year."
So you have your work cut out for you.
Well, yes, it's a kind of job security, I suppose.