Good Data Governance For the Long Term
Information as a Strategic Asset
Customer Ownership Requires Commitment and Metrics
With Web 2.1, We’ll Get it Right
Why Businesses Need to Own Their Customers
Paul Barth Speaks at DAMA-NCR Symposium on "Leveraging Information Asset Management"
Paul Barth on Information as a Strategic Asset


Good Data Governance For the Long Term
By Paul Bergamo, Senior Associate, NewVantage Partners

I define data governance as the management of data to insure its usability, security, integrity and availability. Good data governance specifies a decision-making, ownership rights and accountability framework to encourage the desired behavior. Data governance is as a tool that empowers while maintaining control.

• Empowerment provides clear accountability and roles that allow teams to move aggressively to meet goals.
• Control ensures that data strategy and business strategy are in alignment and serve as a guiding principle.

I’ve observed that most Fortune 1000 companies have a data governance initiative planned, underway or implemented to some degree.

Compliance demands, data disasters, bad business decisions stemming from bad data and the like have been powerful motivators. Most companies are now faced with the need to continually manage data governance and evolve it as capabilities and needs change and mature.

However, I have also seen a number of companies roll over their data governance projects from year to year. They see the potential benefits, but are stopped either by a lack of business support, shortage of IT skills, or more importantly, a lack of understanding of their data. We’ll discuss how to deal with these issues at another time. For now, I’d like to focus on sustaining existing data governance efforts.

I have seen and helped implement all kinds of data governance initiatives. Most fall into one of three categories: Enterprise, Federated and Decentralized. All have their place depending upon the priority, pace, data quality issues, requirements and operational capabilities of the business.

Enterprise

In this implementation, a single central body provides the rules, oversight and day-to-day management of all data assets for all lines of business. These implementations require the most senior management involvement on a day-to-day basis. They involve central committees that oversee processes, approve data decisions and keep moving the initiative forward: They make all the key data decisions.

This approach tends to lose traction over time; it should be seen only as a temporary implementation strategy to gain quick and solid traction. I have implemented a few of these and found them very effective as long as business and IT are clear on roles and responsibilities, operating policies and procedures are clear and accountability is defined. This is a very resource-intensive model that is best used when you have complex data management problems coupled with an urgent need to resolve them.

Federated

This implementation is a hybrid of the Enterprise and Decentralized methods. It couples overarching central oversight and management with local/line-of-business execution and decision-making. In this model, the central oversight/management role is to establish policies and procedures, set objectives, define roles & responsibilities and measure and drive progress. It usually does so by delegating execution. Day-to-day decisions are made locally as ‘local management’ provides its own processes, oversight and measurement. These groups must follow guiding principles and report to a central body but also have direct authority over some assets. They must be accountable to central body for other assets.

This is the implementation model to which most organizations naturally evolve as they seek to find a balance between strong, central oversight and flexible, local decision-making. This is the most sustainable and best approach to ensuring that long-term data management goals are met. I have always believed in a measured approach to technology solutions. I like this model because it leverages the skills of the organization and allows for oversight at a central level without slowing down projects or initiatives unnecessarily because of local decision-making and execution.

Decentralized

In this approach, companies define the rules by which data is governed. Sometimes this is done centrally and sometimes in lines of business as units embark upon their own data governance efforts. In this implementation, all decisions are made locally. Business areas initially like this method because it appears to give them control; in fact, though, it actually only provides short-term relief. We all know how these work. Each project team makes its own data decisions and executes based on its needs without necessarily factoring in upstream or downstream considerations. I’ve never been a fan of this approach. It’s pretty much business as usual and it always leads back to the state in which businesses found themselves before they started their data governance and data management efforts: out-of-compliance and with bad/inaccurate data , untrustworthy data, lack of ownership and so on.

Data governance projects usually starts off well, but without active oversight and strategic positioning, then tend to degrade over time. Think about it. You have a new project with new data to capture. There are clear sponsors, roles and decision-makers. Data is best at this point. But what happens over time? New developers come along, new business leaders take on efforts, or other efforts try to “bolt on” to the previous work. Budget, resource availability, poor processes and the like intervene, and before you know it, duplicate data stores are created or fields in databases are reused for other purposes with limited exposure to the rest of the organization. The result over time: frustrated business leaders who can’t depend on the data they need to make critical business decisions. We’ve all seen it and we’ve all been pressed with cleaning up these problems. In fact, this is probably what led many of us to initiate a data governance effort in the fist place.

Enterprise models tend to break down as discussed earlier. They are great for rallying resources and driving fast and meaningful change, but as organization gain capabilities and address their most pressing needs a plan to evolve the operating governance model must be put in place. I’ve seen organizations try to use this model for too long. Inevitably, they almost always shift to a decentralized approach. Usually, business leaders give up first. They’re followed by the most senior IT executives and eventually by all other key decision-makers. Why? Because there are a lot of pressing needs in an organization and heavy day-to-day involvement in this kind of operational practice inevitably comes in conflict with other critical activities. In addition, projects feel the pinch as all data decisions are assessed and managed through this central group. Planned evolution is critical.

Federated models degrade when their care and feeding is not attended to. Since this is the implementation model to which I believe organizations naturally evolve, it’s important that management take its role seriously. Policies, processes and objectives must be clearly established, success measures defined and progress tracked. In larger organizations, a critical element to sustain data governance is having a small team that can help do the detailed work for the central oversight team while also supporting local teams in their efforts to make informed decisions and communicate effectively across organizational boundaries. Lastly, local teams must not be passive. They must demand compliance within their own borders and in other groups. Non-compliance issues should be raised quickly and non-judgmentally so that they can be quickly resolved.

Decentralized approaches fail for obvious reasons. For one, there is lack of ownership. Downstream business processes cannot depend on data. Unfortunately, this often isn’t noticed until something bad happens. Usually, it’s a critical business fact that is wrong, which leads to a bad business decision. In decomposing the process to find root causes, the bad data issue often reveals itself. Another common symptom is resource overload as duplicate data perpetuates itself in the organization and business and IT resources struggle to determine which data sources are the best and lowest risk.

Good data governance doesn’t involve a lot of cost or complexity. It does require discipline and strategy. In my next post, I’ll outline some success factors.

The idea that information can be a competitive weapon is not new. Yet paradoxically, many organizations today may actually have less control over their information than ever before. Let’s look at why this predicament exists and what CIO’s and business line executives can do about it.

One of the challenge facing businesses and CIO’s is a crushing data quality problem. Typical of this problem is the example of the CFO who receives financial reports from sales and operations that don’t reconcile. This is often because the legacy systems and databases that were developed during the 1980s and 90s are not integrated with new systems and databases resulting from mergers, acquisitions, new products, and new information sources, including web applications. Data integration and data consistency is often minimal or non existent. Data lineage is often undetectable. Without reliable information, business executives are hampered in their ability to make informed, fact-based decisions.

Data quality problems are a chronic and worsening issue in corporate America today. PriceWaterhouseCoopers has reported that 75% of Fortune 500 companies have significant problems because of disparate data.

This has been borne out in our own experience advising many leading Fortune 1000 companies. One firm conducted an audit and discovered that it had 500 Microsoft Access databases in use, each an island unto itself. Another client found that it was processing 28,000 database extracts each night. These copies were being disseminated across the organization without thought to synchronization or control.

What is the impact of bad data? Information quality problems that led to financial restatements cost one public company hundreds of millions of dollars in lost market capitalization. In another case, a large telecommunications firms was able to identify tens of millions in “revenue leakage” attributable to poor data management.

What is the benefit of correct and actionable information? A leading national bank was able to realize tens of millions of dollars in annual savings by making intelligent consolidation decisions based upon integrated customer data. Another financial services firm used analytics to profile its customers and identify cross-selling opportunities that resulted in several hundred million dollars in additional annual revenues without the need for new products.

These examples illustrate the dramatic impact of data quality on business operations. But data quality impacts the customer as well.

Most of us are familiar with the experience of navigating through a customer service organization, only to be asked to repeat the same information by each representative. This frustrating experience occurs because different parts of the organization each believe that they “own” the customer. Often, these operations often have their own software applications and their own copies are of customer data. Databases are populated via extracts from the central data store, and those copies quickly get out of synch with each other. No one has responsibility for data quality or it is treated as an afterthought. Over time, these inconsistencies can become dramatic, resulting in dissatisfied customers and loss of business.

Ironically, the proliferation of new business applications has only made this problem worse. As business units bring in new systems and solutions to address point problems or opportunities, they may also create more data extracts and greater potential for inconsistency. In an unintended way, technology may actually be contributing to the problem of data proliferation and inconsistency.

Addressing these issues doesn’t always require new technology or significant new investment. Improvement begins with the recognition that data is a strategic asset. Organizational alignment and the establishment of business processes that ensure data consistency and integrity on a going forward basis are a necessary first step.

CIO’s have a vested interest in helping their companies recognize that information is a strategic asset, like other critical business assets. For CIO’s who have often expressed the desire for a “seat at the table” in formulating business strategy, the opportunity to take the lead in helping organizations develop an information strategy represents a path from the operational role of “Chief Infrastructure Officer” to the business role of “Chief Intelligence Officer”.

Our series of upcoming postings will explore in greater depth some of the aspects of this topic, including: Data Proliferation, Information Entropy, Data Lineage, Enterprise Data Management, and the Role of the CIO as Chief Intelligence Officer.

Before determining who owns the customer, you need to create an organizational commitment to ownership. This means making an investment in understanding and creating an organization around the customer experience.

Customer ownership is an indication of a mature organization. It doesn’t come quickly.

There’s no logical place in an organization for the customer relationship to live. The sales organization actually tends to be a poor customer steward because of its focus on short-term compensation and sales opportunities.

Marketing is a good candidate, but only if it’s focused on something more than just branding and message. Marketing must be considered strategic and core to the organization.

One client successfully created a Customer Knowledge division. This was an adjunct to marketing that had both a technical capability and information about customers. The group essentially owned the customer data infrastructure and became a trusted source of information about the customer. When the rest of the organization wanted to know what kind of product the customer would buy, they would go to this group. The division had the organizational buy-in, the information and the ability to show how the integrated customer experience affected the organization.

Where does this show up on the bottom line? Ultimately, in customer profitability. It’s the ability to identify which customers are most likely to provide the longest and most profitable relationship with your firm and to make sure they receive the requisite amount of service. That may mean discounts, special service, access to key personnel and other benefits. The organization must be focused upon retaining that customer as a profitable customer.

Conversely, ownership also means being able to identify those customers that will never become profitable. To be blunt, you need to be able to fire some of your customers. This is never easy to do, but a customer that drags on your bottom line does nobody any good.

Only by establishing clear ownership guidelines and gathering and interpreting relevant information can businesses determine where to focus their customer-facing activities. The impact on the bottom line is rewarding.