Part 2 of a six-part series on the "New Age of Master Data Management"
Master data management is entering an 'evolutionary' era where MDM implementations better mirror how businesses use data for numerous purposes, and where people, practices and processes come to the forefront as essential elements. MDM functioned as yet another solution space that was harmed by disproportionate attention to its underlying technologies. But MDM has always been far more about business objectives and needs, as well as the problems to be solved to achieve them. Along with a more overt "de-emphasis" on technologies, we are now seeing more vendors enabling the critical integration of technologies, people, practices and processes with near term and long term business objectives, to achieve the successful utilization of MDM programs.
A major change to MDM approaches calls for combining technology and non-technology paradigms, to reflect the new reality that the business world isn't just "relational". Relationship (network / graph) data continues to grow in importance – but traditional row-and-column databases are inadequate to properly handle such data. MDM approaches and technologies must now process complex connections, hierarchies and relationships between multiple entities / domains. Organizations want to go beyond traditional transactional data to capture data that describes behavioral aspects that can amplify context.
Out with the Old
Over the years, organizations often took wrong turns with MDM by relying on problematic approaches like "big bang" or "bottom up". Either way has usually resulted in big problems and poor outcomes. And both of these approaches have had a history of focusing too much on IT driving the MDM initiative, while failing to incorporate the voices of essential business stakeholders. These traditional approaches usually take too much time, carry too much risk, and end up being too expensive -- often not even achieving essential objectives.
Successful MDM projects require strategic vision that connects to current and future business needs and objectives. But too often organizations have decided that 'strategic vision' means tackling a large scope of work ("big bang"), instead of fleshing out the right strategies before starting any work. Organizations that attempt to execute an entire MDM strategy at once encounter significant obstacles including serious cost overruns, project management turmoil, and chaotic use of IT resources.
The "bottom up" approach relies on IT and / or lower level managers to drive the MDM effort, while disregarding vital upper management and business stakeholders. Often disconnected teams work in silos, but are expected to somehow arrive at the same "desired end state". This introduces the near-impossible task of trying to orchestrate non-collaborative work.
New Approaches Abound
Obviously new approaches for MDM are made better, or even possible, because of newer technologies. But MDM technology by itself will fail to achieve what businesses need. Of primary importance is the work done to establish the right vision and conceptual framework for what master data can accomplish for the business. Organizations need to develop hierarchies of strategies, plan processes and practices, make decisions for design and modeling – among the many preparative tasks that usually ensure more effective utilization of MDM. Then organizations can build out from there, ultimately considering which technologies should be implemented.
Offerings from new MDM vendors and fresh approaches by established vendors are responding to what organizations need and want from master data: centralized reliable data that reflects how organizations work and the kinds of data they need, increasingly in real time contexts. MDM is moving towards faster implementations, broader capabilities in a single extensible platform, and the ability to work with any kind of data.
Organizations are encouraged to explore new MDM approaches and software solutions that are designed to overcome the traditional challenges that have plagued MDM initiatives and that will better fit on-going business objectives, needs and functions:
The Cloud and MDM-as-a-Service
Obviously, SaaS and cloud services have distinctly influenced how customers buy and access software, and how most software solutions are now deployed. Master data management is no exception. As with other software solutions moving to the cloud, MDM can become more user-oriented and fully focused on the real purpose of MDM, without a lot of the technology "baggage" that comes with on premises solutions. Better partnerships between IT and business roles can be forged to optimize MDM processes and how they contribute to real business needs and value.
There are still some obstacles to fully delivering MDM-as-a-Service to business users. But, as with other data management solutions that have moved to the cloud, a great deal of progress has been made. Self-service models contribute to the expansion of different kinds of access as appropriate for different kinds of users. For example, a number of cloud integration platforms provide seemingly simple graphical interfaces that can accelerate usage development and ensure better runtime management. Even technology-adept data management experts want self-service access to MDM solutions. Point-and-click graphical interfaces accelerate all aspects of creating and maintaining MDM processes and applications.
Business-centric Platforms to Solve Business Problems Faster
New and established MDM vendors are working to revamp the MDM design approach. One business-centric approach is based on starting with the desired end state, and all that is required to define it. Then teams work backwards from the desired end state in a reverse-engineering process to figure out how to attain it. The MDM solution platform helps the organization create the models that are needed for a particular end result, and then automatically generates the processes that will carry out the appropriate MDM tasks as extensions of the models that were created. Most "end results" are sharply focused on business needs and value, particularly since extensive business knowledge is required to accelerate development of the resulting MDM processes.
Often this approach brings about a shorter, automated and rule-driven lifecycle that means more iterative development and much shorter implementation times. Business and IT teams can work in parallel, while utilizing the MDM platform to orchestrate and synchronize their efforts. A further extension of such business-oriented platforms leads to simplifying the development of industry-specific business applications that take advantage of master data to solve business problems.
Multi-Dimensional Data – Complex Relationships and Hierarchies
Contextual MDM solutions handle new approaches for multi-dimensional and complex hierarchical data, including social and commercial graphs that underlie business use cases such as customer-responsive interactions across digital channels. Such solutions allow unlimited combinations of categories, context and definitions. Multi-dimensional often means multi-view, reflecting the need to view domains or entities (such as customers) in different ways, in order to derive better insights from data. Today, entities extend beyond the traditional customer and product pairing to include items such as asset, location, supplier, finance, and personnel data.
To better serve business needs, contextual master data management can provide a consistent and multidimensional profile across data and business silos. To meet this imperative, the emphasis falls on the orchestration of master data and processes, to customize a trusted, relevant master view aligned with data consumption needs. The growing interest in multi-dimensional models of master data reflects the huge growth in data elements that are needed by most organizations, as well as increased dimensionality and deeper levels of data.
Contextual MDM utilizes graph databases for additional flexibility in managing multiple domains as "one", instead of creating separate domain models. This approach takes MDM implementations from systems to views. Graph databases can help organizations use master data in more analytic and insightful ways, as well as take of advantage of real-time graph analytics.
Rise of Analytical MDM
With the increase in high performance computing, particularly in support of advanced analytics, analytical MDM now occupies an important position for many organizations. Vendors are integrating MDM capabilities with analytics platforms, for more immediate insight benefits. Underlying technologies include capabilities for machine learning, processing big data, graph databases, and visualization tools. Applications are extensive: fraud detection (real-time and predictive); real-time customer intelligence and product recommendations; reducing costs for many aspects of healthcare.
Unlike traditional approaches for setting up data warehouses for business intelligence (BI) purposes, analytical MDM can perform all MDM functions to ensure data quality, useful data enrichment and the addition of context through consolidated "views" of customer, product and other data domains. Today's MDM hubs often accommodate up to tens of thousands of data elements for multi-domain approaches. Numerous MDM architectures can manage multi-dimensional data with complex hierarchies. Such data provides the wide array of metadata and attributes that are necessary for real-time business processes, particularly for initiatives focused on customer experience and systems of interaction. Analytical MDM approaches (and technologies) are seen as stronger solutions for these sophisticated business needs.
Knowledge-Based MDM – and Taxonomy
An MDM initiative faces a tremendous challenge to align disparate data sources and then perform accurate data quality processes to derive reliable and usable data. Part of the challenge is verifying the validity of data quality results.
Traditional table-driven approaches for data quality needs can be arduous and time-consuming. Organizations have adopted knowledge-based approaches that consolidate information from all records that are accessed in conjunction with an MDM system in order to provide context in terms of customer and other domain-related data. Such knowledge bases often include millions of entries that provide correct and incorrect spellings, relevant patterns, and other data quality attributes. Proponents of knowledge-based MDM claim rapid processing of entire data sets for data quality purposes.
Knowledge graphs bring agility to visualizing master data models and linkages. Conversely, changes can flow into master data and data models through the knowledge graph. Knowledge graphs are built for known needs; which means they can accurately align with business needs. Salient entities and relationships, concepts and content can be incorporated into knowledge graphs.
A frequent partner for knowledge graphs is taxonomy which brings in semantic architecture. Until recently, taxonomy has been used more frequently for enterprise content and information management, and enterprise search. Taxonomy helps MDM data quality processes better handle ambiguity through inherent capabilities for understanding context, resulting in precise meaning. Since many of the underlying processes for MDM and for enterprise taxonomy development are similar, it can be effective to run these efforts concurrently or use one to jumpstart the other.
Re-Thinking MDM with Integrated Approaches
One of the most important changes to master data management approaches has grown out of the fact that much of MDM and data governance revolve around business use cases and business users. Everything done with MDM is first and foremost about the business. Ironically - and painfully - the increased involvement of different business roles constitutes a 'new' approach for many organizations.
Approaches that can help the "people issues" of MDM programs include:
- Embedded collaboration practices and technologies – integrating business and technical users to achieve intelligent and optimal usage of MDM
- Business user access and visibility into appropriate MDM processes, including continuous reviews of data and use of tools like data profiling
- Make it easier and faster for everyone to do their share of the work
- Comprehensive and responsive orchestration of many different teams and processes, to work as one
While new approaches have been discussed as individual items in this article, for real world implementations, multiple aspects of different approaches usually come into play for a particular organization, and need to be well integrated for effectiveness. Once again, such integration reflects the complex world of business.
In the past few years, new vendors have entered the MDM space with solutions that have a better focus on what businesses need. New and old vendors are also dismantling the old ways of how MDM has been implemented to more realistically support the diverse array of data sources that businesses need to utilize, particularly for faster, more agile analytics processes. The next article of this series will examine the evolution of MDM technologies.
Image source: landarchs.com
About the author: Julie is an independent consultant and industry analyst for B2B software solutions, providing services to vendors to improve strategies for customers and target markets, products and solutions, and future direction. Julie has expertise in several solution spaces including: data integration and data quality; analytics and BI; business process, workflow and collaboration; digital marketing, WCM and social media; and the pivotal importance of user and customer experiences. Julie shares her takes on the software industry via her blog Highly Competitive and on Twitter: @juliebhunt For more information: Julie Hunt Consulting – Strategies for B2B Software Solutions: Working from the Customer Perspective
