Simplifying Master Data Management Deployments by Marty Moseley, CTO, Initiate Systems Compared to the myriad group of “integrated” systems that most companies are managing today, master data management (MDM) solutions are much simpler to manage and maintain, and provide companies with more business benefits. Unfortunately, MDM technology is developing a reputation for being complicated and taking a long time to implement when the reality is that the process can be dramatically simplified if companies plan before they implement. To streamline the implementation process, companies need to make a number of decisions about data, business processes and technologies, before an MDM project begins. First, they need to make critical decisions about what data to master and why. Then, they need to address other “typical” business process issues, common to any IT program, including building out the business case; getting buy in and budget approval; figuring out the business process, strategy, enterprise architecture, rules, policies and procedures; and dealing with change management. Finally, they need to make technology decisions, including selecting the best technology to match data and business goals and ensuring the choice is simple, predictable, low risk, and can be implemented in months, not years. What Do You Master and What Does It Mean? The first step for a company considering MDM is to understand the benefits of an MDM solution and how it differs from the way things are done today. MDM solutions generate and maintain an enterprise-wide “system of record” that contains the consistent, reliable information necessary to perform vital business functions across a large organization. MDM deployments result in a massive simplification of the widely distributed, uncoordinated data management solutions that most companies struggle with. The benefits of MDM include enhanced revenue and profit, improved customer service, lowered operational costs, easier compliance, managed risk and better strategic decision making and business agility. Once a company knows it needs MDM, how does it decide what to master? The most basic advice here is to pick the data that will deliver the biggest bang for the buck first. The best way to decide which data to choose is to identify the largest areas of pain within an organization. Decision makers should take a look at areas where costs and data defects are out of control, customer satisfaction is trending downward, inconsistent pricing exists across channels, market share is shrinking, customers are complaining about marketing, regulatory requirements are creating a stranglehold, reporting to Wall Street is painful, or there are significant untapped opportunities that could be capitalized on. A company should choose the area that will deliver the greatest measurable return and tackle that first. The process of determining the first project will most likely make it clear what the other top five or 10 projects might be. After a company makes the decision about which data to master, it is important that it understand what is entailed in the process. An MDM solution puts all the myriad data sources into a single complete database to create a central, single version of the truth for that data domain. Once data is centralized, duplicate records will be resolved, relationships between data will be detected and declared, and data will be made available to the applications, people and processes that need it. Different business uses and security restrictions require that not everyone has the same access rights or ability to view data, so MDM technologies need to be able to control access, enforce security policies, and provide logging and reporting on details. MDM implementations can be onerous and complex if the team does not focus on simplifying each step and its overall approach to the MDM project. Below are specific business and technology guidelines that will help to simplify MDM deployments. Five Key Lessons to Ease the Pain It’s important to first consider five key lessons to help ease the pain and streamline MDM business requirements: Look for simple solutions to business problems. Trust your instincts. If the proposal sounds as if it will require years to design, implement and integrate, you’re probably right. To simplify, manage scope. Don’t try to master all data in one project, pick the one type of data that delivers the biggest bang for the buck and fix that first. Get value quickly. As always, time is the enemy of business, so rapid implementation times do matter. If you can’t get a system into production within six-to-eight months, it is taking too long. Save money through automation. The more MDM technology does to find and maintain data relationships and resolve data quality issues automatically, the fewer people that are required to maintain data accuracy and the more money the company saves. For example, today it is not uncommon for companies to have multiple copies of data quality software to manage multiple systems. When the data quality function is housed within the MDM solution it only needs to be mastered once, so a company saves money by not having to manage data quality in multiple places, normalize business rules across software from multiple vendors, or purchase multiple copies of data quality software to maintain different systems. Make search capabilities a priority. Robust and sophisticated master data search capabilities make it much easier for point-of-service employees to locate accurate files. Built-in search capabilities enable MDM solutions to decrease the number of duplicate records created and ultimately lower downstream data stewardship and data quality costs. Try not to over-invest in technology. While it’s important to find an MDM solution that is agile enough to meet your current and anticipated business needs, avoid the temptation to buy a solution for a problem that you don’t have. By managing scope and avoiding complexity, you should be able to clearly define your technology requirements and avoid spending more than necessary. From a Technology Perspective If you want to simplify MDM, do not undertake a single comprehensive project that will take several years to implement. Instead break the MDM project into segments and deploy each individually. Technology should help make the process of implementing MDM easier, not add to its complexity. Here are five technical guidelines that will help most businesses further simplify their MDM program: Buy what you can, build what you must. Many companies significantly underestimate the difficulty in writing their own matching engine, or they try to manage reference data with a simple list of valid values. With matching engines, you can quickly get into a very brittle deterministic rule set that is unmanageable, inflexible and non-extensible, and which performs poorly. Is it a core competency of your company to build a proprietary MDM hub? Probably not. When you can, for less time and money, buy one. Remember, building the rules for an MDM system is only one part of the project, you also have to maintain and change the rules as your data set grows, you bring on numerous systems, or you need to manage conflicting rules from different constituents. Insist on architectural flexibility and adaptability. Technology is only as good as its ability to adapt, so it’s important that an MDM solution has architectural flexibility and a “roadmap” of deployment styles. Ensure that your MDM solution can automatically adapt to your data as it changes. A flexible MDM offering will enable you to easily add sources without modifying rules, and will maintain high accuracy and performance levels as you add more data. Start with a registry approach. A registry approach to MDM matches and links data from disparate systems to provide a single customer view without requiring organizations to build a centralized data repository. Registry-style MDM solutions are often easier to deploy and generally deliver the fastest returns. If your MDM technology requires months and months of setup before you see a return, chances are you’ve bitten off too much. Leverage data federation wherever possible. Federation creates a single version of the truth for data by resolving heterogeneous instances into a single, trusted view. MDM solutions with built-in data federation reduce the complexity of deploying federated solutions. Consider the possibility of data exchanges. The future of MDM for many organizations will include connecting and sharing data across multiple organizations and agencies. With exchanges, every organization doesn’t have access to every piece of data. What is required is the ability to securely share critical data across organizations, so it is important to select an MDM solution designed to easily support sharing and accessing data, while protecting privacy. Demand accuracy even while scaling. It is critical to choose an MDM solution that is able to maintain accuracy while performing and scaling, no matter how much data you are managing. Performance and scalability matter even for smaller data sets, but as the amount of data grows, performance and scalability demands will increase and so can hardware procurement and management costs. It is important to understand your exposure and ensure that accuracy will not be compromised. Don’t reinvent the wheel. You don’t have to spend a lot of time and energy redesigning something that has been done well by someone else. Resist the temptation to spend countless person months designing the perfect data model in house. Instead, choose an MDM vendor that has lots of experience developing and perfecting these models, so you don’t have to. Find a partner that has “range". MDM technologies are relatively new, so be sure the MDM partner you choose has experience solving a wide range of data problems and won’t have to reengineer your systems. Be skeptical of vendors that want to start at a departmental level, but promise to grow to an enterprise level, as often these technologies cannot scale easily. Getting the MDM Solution You Need Simply and Easily MDM can provide companies with many business benefits and do not have to be difficult to deploy. The key is to approach the project in pieces and plan before you implement. Once you have considered all of your organization’s data, business and technology questions and have made the critical decisions, the MDM implementation will be much simpler. A company that takes this approach, and starts today, should see results within six or eight months. About the Author Marty Moseley is a 25-year IT industry veteran with extensive systems architecture experience. Moseley is an accomplished speaker and author on technology topics including data governance, customer data integration (CDI), master data management (MDM), service-oriented architecture (SOA), software architecture and product-line architecture. Moseley currently serves as chief technology officer at Initiate Systems, a leading provider of master data management software for companies, healthcare organizations and government agencies that want to create the most complete, real-time views of people, households and organizations from data dispersed across multiple application systems and databases. He can be reached at mmoseley@InitiateSystems.com and additional information on Initiate Systems is available at www.InitiateSystems.com. About Initiate Systems Established in 1995 as Madison Information Technologies, the company became Initiate Systems, Inc. in 2003. The company has 175+ software customers across numerous industries. It is cited as a "visionary' by Gartner and a “leader ” by Forrester. Initiate Systems operates globally through its subsidiaries, with corporate headquarters in Chicago and offices across the U.S., and in Toronto, London and Sydney. Citation Moseley, M., "Simplifying Master Data Management Deployments", DSSResources.COM, 01/09/2009. Kevin Johnson, Manager, TechImage Media Relations (for Initiate Systems), provided permission to archive and feature this article at DSSResources.COM on Decemebr 2, 2008. This article was posted at DSSResources.COM on January 9, 2009.