Data Quality Considerations in Master Data Management Structures

Data Quality Considerations in Master Data Management Structures W H I T E PA P E R : DATA QUALITY Michael Overturf Navin Sharma WHITE PAPER: DATA QUALITY Data Quality Considerations in Master Data Management Structures 2 OVERVIEW COMPANIES ACQUIRING COMPANIES. HUMAN RESOURCES SHARING INFORMATION WITH FINANCE. BUSINESSES SPANNING MULTIPLE COUNTRIES. WHAT DO ALL OF THESE SCENARIOS HAVE IN COMMON? THE SHARING OF DATA. WHAT IS THE CRITICAL NEED SHARED BY ALL OF THESE SCENARIOS? DATA MANAGEMENT. MASTER DATA MANAGEMENT, TO BE EXACT. IN TODAY’S BUSINESS ENVIRONMENT, MERGERS AND ACQUISITIONS, DATA SHARING, AND GLOBAL MARKET PRESENCE ARE COMMONPLACE. TO BE SUCCESSFUL, COMPANIES MUST NOT ONLY GAIN CONTROL OF, BUT MUST DRASTICALLY IMPROVE THEIR CUSTOMER DATA. TO DO THIS, IT IS PARAMOUNT FOR COMPANIES TO IMPLEMENT AND ENFORCE ENTERPRISE-WIDE MASTER DATA MANAGEMENT SOLUTIONS. www.pbinsight.com CORRECT CUSTOMER INFORMATION IS THE LIFEBLOOD OF EVERY COMPANY What is Master Data Management? recent Ambysoft survey1 of over 1,100 US IT organizations, 3 only 40% of companies had some sort of developer Master Data Management (MDM) is a process for effectively regression test suite in place to validate data, yet over creating a complete data context view in medium to large 95% affirmed their belief that data is a corporate asset. distributed data environments. MDM may be accompanied Two thirds (66%) of those respondents stated that their by Customer Data Integration (CDI) systems that provide a development groups would sometimes bypass their data simplified access layer to corporate applications. The interplay management staff, if they had one, because they were ‘too between MDM and CDI provides system architects the facilities slow’, ‘too difficult’, or ‘offered too little value’. to ensure the operational contiguity of data. More simply put, it is the ability to understand the transactional context between a In a more recent 2009 global survey of businesses corporation and its customers. conducted by the Information Difference, one third of the respondents rated their data quality as poor at best and However, while most business managers agree on the only 4 percent as excellent. Only 37 percent indicated that value of this knowledge, many lag in the adoption of they currently have some form of data quality initiative in requisite technologies to attain it. This paper discusses the place, and surprisingly 17 percent said they have “no plans” continued intransigence of corporate data quality problems at all to start a data quality initiative. The two top barriers and how such data quality issues can be resolved within the to adopting data quality were: (1) Management does not see MDM and CDI implementation context. this as an imperative and (2) It’s very difficult to present a business case. State of the Art Data Quality Management Most system architects know why data quality is important. What can operational IT managers, data, and system Still, data management as a discipline is surprisingly architects do to improve quality against this background? underdeveloped in many companies. For example, in a Other Internal Clients Data Entry Outsourced Procesing Logistics Finance CORPORATE NETWORK Direct Marketing DWH External Data Provider CRM Other Division Other Business Typical IT environment that obviates the need for MDM 1 Scott Ambler’s August 2006 ‘Current State of Data Management’ Survey 2 The Information Difference, July 2009 “The State of Data Quality Today” Survey WHITE PAPER: DATA QUALITY Data Quality Considerations in Master Data Management Structures 4 Architectural Challenges use, allowing the option to leave the source data as is and defining DWH entries as entire or partial master records. Mid-to-large corporations have installed a host of A Hybrid approach (data-less access or zero-latency data specialized information systems that support various warehouses) incorporates components of both schemes. In parts of the business. A corporate network could include all approaches, data quality and metadata management are a CRM system from Oracle, a transaction management often done parenthetically, and should be executed by a system from IBM, a logistics management system from SAP, high performance, extremely flexible and easy-to-use data and a financial management system from Lawson – or any quality system. possible combination of the listed systems. Data Quality Functions in MDM Each system has its own data definitions. For example, a CRM system may list customer name as [<First>, A salient feature of a customer-based MDM system is a <Middle>, <Last>], while a legacy system will have an unique customer identifier that maps all relevant structured aggregated [<Name>] field, listed as [<Name1>, Name2>, customer data. This customer data may include personal <Name3>]. When attempting to combine this data into information (name, address, phone, email, etc.), business a comprehensive understanding of each customer, a and consumer demographics, communication preferences, programmer or database designer must arbitrate between order history (product, location, reclamations, satisfaction many different interpretations and encodings of essentially indicators, etc.), relevant financial support data for credit/ the same, or similar, semantics. This can generate underwriting, and contact management history. thousands of mappings and combinations. Without a manageable data quality rules database, this becomes a Fully implemented MDM is an important feature of daunting situation. Customer Communication Management (CCM). CCM data is broadly categorized into three major groups: Customer Variant update frequencies of each individual data system Intelligence, Channel Intelligence, and Production add further complexity. Online Transaction Processing Intelligence. Customer Intelligence manages customer data (OLTP) systems continually update their data, but metadata more effectively. Channel Intelligence breaks down delivery updates are usually marked with undesirable service channel barriers for more consistent communication. interruptions, as reflected in the above surveys. Cyclical Production Intelligence transforms data for more effective systems, such as financial or sales management systems, message delivery. Customer Communications Management update in different contexts from different sources. MDM describes the integration of structured customer data, unstructured content, customer emails, documents, or Master database taxonomical relationships with their data messages, and structured production control data. sources have three primary forms: a federated architecture, a data warehouse (DWH), and a hybrid. A federated MDM supports closed-loop CCM by relating a customer architecture leaves the source data in its original location response (e.g.