Mastering the Data Management Maze

The ability to merge all the disparate, oft-conflicting records you have on customers and transactions into one authenticated master file is without doubt appealing, but what challenges does attaining this data nirvana present, and how can MDM help address these issues?

With more than 30 years of IT experience, Deloitte Consulting’s Jane Griffin is in an ideal position to moderate this issue’s MDM roundtable discussion. Throughout her career, she has developed and led business intelligence and data warehousing practices in a number of leading professional services firms, including one she founded in 1986, and has assisted and advised multi-industry clients in designing, developing and implementing technology and processes to leverage information. Jane serves on several partner advisory boards for data management vendors and maintains high-level relationships with leading vendors in the business intelligence and data management sectors, authors a monthly column in Data Management Review and has also published more than 150 articles in various trade journals. In addition, she participates in frequent international speaking engagements on topics such as customer relationship management, data warehousing, repositories, data modeling data quality, consensus building and enterprise-wide business modeling.

Here, she discusses the finer points of master data management with a number of leading providers:

Len Dubois is the Vice President of Marketing for the Trillium Software division of Harte-Hanks LLC. Dubois has been with Harte-Hanks for seven years and has more than 15 years’ experience selling and marketing high-tech solutions.

Howard Dresner is Chief Strategy Officer at business performance management provider Hyperion. A preeminent authority on business intelligence, Dresner joined Hyperion from Gartner where he was formerly Vice President and Research Fellow.

Sunil Gupta is Director of SAP solution marketing. He has been working in the enterprise software applications area for the past 16 years and represents the Product and Technology Group at SAP with a specific focus on master data management.

BMUS. IQ Matters, a recent survey conducted by CFO Research Services in conjunction with Deloitte Consulting LLP, found that fewer than half of senior IT and finance executives surveyed believe they have achieved their information quality objectives, and 61 percent believe they could do a better job of providing financial information that accurately reflects the performance of their companies. Why do you think companies are having such difficulty getting the information they need?
LD. Many senior level managers focus on the IT systems in place and not the data feeding the applications and programs. As a result, they tend to believe that the system has failed and the corresponding investment in the technology has been wasted. Most organizations fail to understand at the outset that you cannot automate a process for better information without ensuring that the information itself is fit for the business purpose you have in mind. For instance, the most complex and expensive ERP application, which ties together sales data from billing systems and customer data from CRM applications, is not going to accurately identify the number of new customers acquired by an organization if the system cannot differentiate a new customer from a preexisting customer. It may simply register the new record as a new customer and create a duplicate record.

SG. I think that the difficulty that companies are having with information quality can be traced to the piecemeal way in which information systems have been planned and implemented in the past – each department, division and region setting up separate applications, defining their own data models and running standalone data stores. When it became clear that departments needed to share information with each other and with trading partners, the silos of information proved difficult to consolidate and share. Organizations are looking to be competitive, compliant and agile in the 21st century; information sharing inside the company’s four walls and understanding that a single consistent view of their business is a prerequisite to these goals.

HD. The difficulty stems from a lack of enterprise-wide data governance programs and processes that treat data as a corporate asset – in particular, master data. Change is constant in every organization, and as such it must be rapidly and accurately reflected in the company’s business performance management systems. Master data is the ‘vehicle’ that allows us to accurately represent our organization in these systems; it is comprised of all the reporting dimensions, hierarchies, business rules, attributes and properties that define our organizations. If we don’t have processes in place to effectively manage change and administer the lifecycle of master data, it will be inconsistent across enterprise systems. This results in lack of confidence in the quality of our information, and drastically increases the risk of non-compliance. It is for this reason information quality is a topic top of mind for executives today.

BMUS. Most leading companies have been investing heavily in IT for the past 40 years. And, they continue to invest enormous amounts on IT to help enable enterprise operations and management capabilities, expecting these investments to provide them with better access to accurate, timely, consistent, and relevant information. Yet, the reality seems to be the opposite. The more that is spent on IT for operations, the less useful generally the information. Please discuss how implementing an information management strategy can help reduce cost and improve ROI.
SG. Investing in technology alone is not the complete answer to meeting information management objectives. To reduce cost of ownership and improve ROI, an organization needs an information management strategy that thinks of the organization as a whole living entity. Just as the body cannot have a separate brain for every limb, companies need an enterprise-level, executive-sponsored data governance model, which functions on both an organizational level and a solution level.

Organizationally, a data governance council mediates across groups to set data standards and practices for the entire company. The technology solution must be capable of executing the council’s mandates. ROI and lower cost of ownership are achieved by implementing the council’s standards, which in turn eliminate redundancy (or prevent it), increase the opportunities for additional sales, and reduce errors based on flawed information (ranging from incomplete product descriptions to financial reporting).

HD. Investment value is not about the number of systems, the volume of data or the number of dashboards and reports our IT departments can generate. It’s about the quality and consistency of enterprise-wide information, and how actionable it is for making decisions. A comprehensive information management strategy must address typical high-cost problems associated with effectively managing master data, ensuring financial and operational data quality and implementing enterprise-wide data integration. Such strategies increase the alignment of business and IT, thereby reducing the costs of re-work to adjust inconsistent reports and analyses. They allow for standardization in enterprise data management technologies, which reduces total cost of ownership. Lower audit fees and compliance costs are also important results. Increased efficiency and time saved by IT personnel, plus the ability to redeploy resources for more productive activities further reduce costs and improve ROI. There are also a variety of intangible or harder to quantify benefits that ultimately improve ROI, such as more timely decision-making, increased accountability, trustworthy information, greater visibility and transparency.

LD. Organizations have only recently begun to fully understand the benefits associated with the implementation of data quality processes, procedures and tool implementations as part of data intensive initiatives. In fact most of the investment has been in hardware and major system software acquisition. Little has been invested to ensure that the data itself is well understood and fit for its intended business purpose. Accurate, complete and consistent data impacts financial results and ROI by driving new revenue from accurately identified prospects, decreasing operational costs by allowing data to be shared across operational units, optimizing data integration processes and reducing failure at the systems level, and creating growth opportunities by directing scarce sales and marketing resources at high value, retainable customer opportunities. Many organizations follow a data quality maturity profile that allows them to scale from limited small project successes to enterprise-wide data quality initiatives that can yield savings in the tens-of-millions of dollars.

BMUS. A leading analyst firm recently reported that most executives surveyed believe data management is just an IT issue. We believe it isn’t. So how does data management impact business issues and performance management?
SG. Businesses depend on information, and information is derived from data; data management highly impacts business and performance management. If the information about any aspect of your operations – whether it is about customers, products, suppliers, financial data or capital assets – are unreliable, then you cannot produce reliable reports, generate reliable forecasts and ultimately you cannot make sound decisions. Ultimately, data is at the heart of all business processes, and bad data results in sub-optimal business processes.

HD. Managing data is also a business issue, since it’s the other areas of the company, not IT, that are the primary consumers of information for making decisions that shape the company’s future. The main objectives of data management are to improve the accuracy, quality and integrity of data. Therefore, data management has direct impact on various performance management issues, such as reducing compliance costs and risks, achieving a single version of the truth for financial and operational data, faster close cycles and greater confidence in reporting and analysis. Effective data management also provides better visibility and insight into the business, resulting in active participation of business users in managing corporate data assets. In an organization with solid data management, users demonstrate improved flexibility and responsiveness to industry or business changes and ultimately improve their ability to make timely decisions.

LD. Poor data quality costs organizations a lot of money and losing money is a business problem, not just an IT problem. For example poor data quality can send a shipment to the wrong address, charge the wrong price for an article of clothing, and discount a sale at an inaccurate rate. Poor data quality can fail to recognize a change to a product number and send the wrong item to a factory for manufacturing purposes. Inventory management systems then recognize that a part has been shipped and reorder a replacement part, creating an overstocked item when the incorrectly shipped item is returned to inventory control. A key differentiator for many organizations that have implemented successful data quality programs is understanding the dividing line between IT, which owns the process of automating data, and business managers who are responsible for identifying the meaning of data and its usefulness for business purposes.

BMUS. Maintaining a competitive advantage in today’s globalized markets requires companies to shift the focus from enterprise operations to enterprise information. Why should companies take an enterprise view for approaching this challenge?
HD. As I mentioned earlier, companies must treat information as a strategic asset. This implies that it is a corporate, top-down initiative that comes from the highest levels of the organization; which is why most organizations today have C-level executives whose focus is information. Taking an enterprise view of information assets allows companies to implement application-independent policies and programs for managing data, and in particular, master data. This is crucial for eliminating silos of information and for accurately and consistently representing our businesses in logical data models, regardless of the applications or data stores that are deployed. An enterprise view of information also forces a focus on process, not just technology. You cannot implement enterprise data management strategies without defining procedures for describing data assets, managing logical data models and changes and defining governing policies, to list just a few. In addition, it’s critical to outline user responsibilities and accountability. Enterprise information management is a business process issue, not a just data or system synchronization issue.

LD. Although an organization’s immediate goals may be served by addressing tactical problems that arise from poor data quality there is no reason this should not be considered as a first step in establishing an enterprise-wide approach to information management. Master data management and data governance, both highly data intensive and strategic in nature, is fast approaching the mandatory stage of most C-level decision maker’s list of top priorities. As a result, many organizations are solving tactical problems today by investing in strategic solutions that will enable them to implement data quality on an enterprise-wide scale. Chief among their reasoning for embarking on this type of solution is the ability to leverage best practices for codifying and standardizing business rules, mandating business user input and sign-off for transforming information that will affect ongoing business processes, and the ability to monitor and audit data for compliance purposes. Lastly, the global nature of business now mandates an enterprise approach to information quality in order to effectively conduct business on a worldwide scale.

SG. Organizations should take an enterprise view so that they do not perpetuate the cycle of inaccurate and incomplete information. When segments of an organization are allowed to develop isolated sets of data, inconsistencies grow and fester. With an enterprise view, the entire organization cultivates the same set of reliable and accurate information, which is shared to various segments. Each segment may choose to augment that enterprise information with other attributes, but the core set of master data is the same. This centralization prevents siloed data stores that can cause problems in all lines of business. Also as organizations move to services oriented architecture, the need for a solid master data foundation becomes even more pronounced since enterprise services require an accurate source of master data.

BMUS. Building a data repository may seem easy. But if information is truly viewed as a company asset, what is required from a governance structure to provide an effective MDM strategy that delivers information quality and value to address this enterprise information shortfall?
HD. An MDM strategy requires the identification of a multi-functional team that has multiple responsibilities, including: defining the lifecycle of master data; identifying initial and long term goals; developing corporate definitions; establishing enterprise-wide policies and business rules for managing changes; defining approval levels and internal controls; and identifying the systems that will interact with the MDM solution, etc. Other key elements of an MDM governance structure are executive sponsorship and participation, as well as business user involvement in maintenance of master data.

The governance structure of an MDM strategy must also allow us to dictate the path to follow for making the MDM solution the central source of record for corporate master data, as well as the source of entry for changes.

LD. Master data management, customer data integration (CDI) and product information management (PIM)) initiatives, developed without concern for operational data quality and well-defined Data Governance practices, are at the same risk of failure as their predecessors in the CRM and ERP space were several years ago. Operational (real-time) data quality processes essentially close the loop on information quality efforts in large system implementations, in effect, preventing the corruption of existing data files and ensuring adherence to established standards. Data Governance teams, mandated and funded by senior management within organizations are established for developing the process by which an organization manages data quality. To ensure success, Data Governance teams must have the commitment of senior management, both in terms of time and funding, to establish a culture of quality. These teams must also have the ability to mandate change in business processes when necessary, and the technology to monitor and audit data to ensure adherence to standards.

SG. Effective data governance requires representation from both IT and each line of business. The group must come to a common understanding (a definition) of what is master data within the context of the organization. It also requires strong executive sponsorship and constant communication to all employees to ensure that the asset is properly maintained. Most of all, a data governance council needs authority. They need the power to effectuate changes. SAP actively participates in and drives data governance strategy initiatives on an ongoing basis, and these are some requirements that customers have shared in discussions with SAP.

BMUS. Having timely, accurate, reliable and consistent information doesn’t just help enable better decision-making, strategic planning, and operational efficiencies—it also greatly impacts a company’s ability to meet regulatory reporting requirements. Please discuss how MDM can improve a company’s ability to meet compliance regulations and help decrease risk.
SG. Compliance requires accuracy. Master data management improves a company’s ability to meet compliance requirements by ensuring a single version of the truth. By removing duplications, incomplete records, taking the best values out of all possibilities, and perhaps even augmenting your data with external sources, an organization can create more reliable master data. This data can then be shared from a central point so that all systems are working from the same core set of information. Master data management can help companies comply with Sarbanes Oxley by providing the structure and the solution to manage financial objects on an enterprise-level. Executives can then accurately report their results. Not only is reliability improved, but the consolidation of this data from disparate sources is automated, which is usually a very manual task for companies who have not implemented MDM. One of SAP’s customers, a very large high tech company, recently had to restate financial statements (and got a new CEO in the process). That customer has implemented SAP to prevent a repeat of that incident.

HD. With an MDM strategy, companies have visibility into their true financial and operational performance, which is crucial for adhering to compliance regulations. MDM enables transparency of all processes related to how a corporation defines its information assets, as well as how it manages changes in its master data due to normal course of business or disruptive events such as mergers or acquisitions. Effective MDM also delivers complete auditability via verifications and validations, plus roll back capabilities to any point in time. It can also audit what-if scenarios for impact assessment and comparisons. Other important elements for regulatory compliance and risk reduction are internal controls and enforcement of business rules, which are achieved through granular security and accountability by user. Finally, MDM eliminates manual, error-prone processes, thereby increasing the certitude and timeliness of reporting and analysis. This in turn reduces close cycles, which is also a concern related to compliance.

LD. A byproduct of an effective MDM initiative is accountability. Failure to comply with regulations increases financial risk and can ultimately destroy a company’s credibility. The ability to accurately identify and flag customers or transactions as individuals or entities identified on government ‘watch-lists’ is a cornerstone function of the data quality component of a MDM solution. In addition, reducing risks associated with identity theft, fraud detection and financial crime are also key features of an enterprise data quality tool. Compliance with and adherence to regulations is here to stay, and organizations must establish a well-documented approach for accountability to these regulations. MDM helps provide the means and access to an enterprise and organizational view of customers and their relationship with the organization, thus establishing the accountability needed to comply with regulations and decrease risk.

BMUS. IT investments should be aimed at supporting the initiatives that are necessary for addressing a company’s information challenges. Yet, it starts with identifying what information is required to run their company. What are the steps IT and corporate executives should consider taking to address this challenge?
LD. There are multiple steps that an organization can take to understand their needs for corporate information. First and foremost is the requirement for defining the long- and short-term business goals for corporate information. Short-term objectives usually relate directly to a specific project while longer-term goals usually take into consideration how the data will effect other operational applications and systems. It is alarming how many organizations fail to consider what other applications will share data processed at the point of entry within an organization. Additionally, organizations should consider the following: what source or sources contain needed information, what is the level of quality within each source, what cleansing and standardization is necessary to meet the requirement, and what problems or anomalies must be addressed as part of this project in order for success to be achieved.

SG. To best determine what information is an asset and what information is just white noise, IT and corporate executives need to roll up their sleeves and get their hands dirty. Executives need to have a vision of what they hope to accomplish. IT can help by asking questions to build data models. I am willing to bet that all executives do not all agree that marketing and sales are the only lines of business that need better information. I bet that the product development team wants more reliable information about the availability of parts and materials. I bet the purchasing manager wants to receive more reliable information from his suppliers. And do you think the COO wants better information about her capital assets in all 46 countries that her company operates in? Once executives analyze the gap, then they can work with IT to model the data. The modeling then enables IT to source an appropriate solution to address all the different data objects. Clearly, the solution must address all the data types so that IT doesn’t waste time and resources stitching together multiple solutions for reach object type.

HD. This is a classic challenge that I’ve been researching and writing about since many years ago when I coined the term ‘business intelligence’. The process should start with an analysis of business users’ needs for actionable information in their daily jobs, from executives down to lower levels in the organization. This usually takes the form of KPIs, analytics, reports, scorecards, plans and budgets. The key question is “what do I need for measuring performance of my area of responsibility,” which must consider both financial and operational measures. What’s important is to compare these needs with the currently available data in the organization and establish the gaps. Often, organizations set forth ambitious programs for measuring performance only to realize later that the state of their data cannot support such programs. This is where MDM comes into play. Unless we have consistent corporate definitions for our information assets; procedures for managing changes to master data; business rules, a clearly defined master data lifecycle, etc. companies will have trouble determining the right information they need to manage – not just run their businesses.

BMUS. What are the pain points driving companyies to face their information requirements and consider an MDM solution?
SG. There are many pain points driving companies to an MDM solution, from gaining a complete view of customers to more efficiently managing assets to reducing compliance exposure. Companies are faced with increasingly tight profit margins that force them to look carefully at all leakage points. So one example of a leakage point is the supply chain, one area where most companies think they can do better. Did you know that according to a study by Global Commerce, 30 percent of supply chain costs are attributed to correcting product data errors? SAP has helped manufacturers, wholesalers and distributors tighten their operations by helping them provide more accurate product, parts and material data within their enterprise, as well as enabling them to share data with trading partners. Improving supply chain transparency also results in improved customer service by providing a clear consistent view of their products across operations. But the supply chain is just one area where MDM is being adapted to solve chronic issues.

LD. The type and number of pain points that an organization experiences are as varied as organizations are themselves. Several of the key requirements that drive an organization to evaluate information quality in conjunction with the implementation of an MDM solution include the need for global data quality (automatic processing of global name and address data against deep, pre-defined rule sets that reflect local cultural, lingual and geographic issues); sophisticated matching and relationship linking (de-duplication as well as sophisticated relationship linking and record classification based on inexact values); tunable rules with traceable results (understanding and defending why results were produced to ensure compliance with regulations); business user ownership and control (enabling business users to manage enterprise rules, divisional rules and local exceptions from a single point of control); and enterprise scalability (one data quality platform for CDI, MDM, supplier data, transaction-processing applications and data warehouses).

HD. There are many reasons that customers adopt MDM as a key element in their information management strategies. These include an inability to obtain a single version of the truth of their corporate information; the high costs and time wasted in reworking inconsistent figures in their reports; a lack of visibility and transparency into their data management processes; high compliance costs and lengthy close cycles; an inability to audit master data changes (and the resulting high risk of non-compliance); difficulties in maintaining historical versions of master data; and the high TCO of in-house developed solutions that only partially address MDM needs.

BMUS. Although delivering data quality might be the most direct way to solve the problem, maintaining those links can quickly become a nightmare. What is the most efficient means to designing for long-term efficiency?
SG. First, data quality is a lifestyle, not an event. An enterprise must design, develop and implement ongoing processes to maintain data quality, therefore the solution must be capable of not just linking data together, but must actually create a master repository that represents the best values for all master data. This system then becomes the best source of truth and feeds clean, reliable master data back to the source systems and to any downstream applications. Second, the solution must not only provide a single place for consolidation, but it must also support updating and creating master data. This means that master data, as well as data quality rules and checks, live in one place and are managed centrally but can be easily shared with other systems as needed. Being able to distribute master data administration in a secure fashion is essential, but having said that there are places where operational systems need to control master data maintenance, such as after a merger, and a good master data management solution will support both deployment scenarios. The third part is mandate by a data governance group that defines how to put this framework in place, resolving duplication but providing correct access to master data across the enterprise.

HD. A key consideration for long-term efficiency of MDM strategies is to design processes with short-term and long-term goals in mind. It’s also important to clearly identify a company’s master data lifecycle, as well as business user involvement in managing changes, manual or batch. Other considerations include the use of service-oriented architectures and implementation of individual export/import/blend profiles by downstream and upstream applications. Most importantly, an MDM strategy is a journey, so adequate flexibility in the associated processes and MDM solution implementation is important as deployments evolve towards incorporating all corporate master data into a central source of record.

LD. According to Nigel Turner, head of ICT Customer Management for BT, the success of any long-term data quality improvement program depends on a few key factors. Every specific improvement effort must begin with rigorous data profiling. Organization’s must provide an official master record for each entity within the organization, and ensure all systems share the same view of entities. They must recognize that successful business processes depend on having quality information – and that data quality is not an end in itself, but rather must be tied to a business case. They should invest in tools, processes, methods and culture to enable information quality, and remember that data quality is a continuous process, cyclical rather than linear in nature.

Finally, it is essential to select objectives and projects appropriate to the data management maturity of the organization and place responsibility for information quality with the business, not with IT.