Data Governance can have various definitions, depending on the audience. To many, Data Governance consists of committee meetings and stewardship roles. To others, it focuses on technical Data Management and controls.
Holistic Data Governance combines both of these aspects, and a robust Data Architecture and associated diagrams can be the “glue” that binds business and IT governance together. Donna Burbank, Managing Director, Global Data Strategy, spoke at the DATAVERSITY® Data Architecture Online Conference about opportunities for quick wins that are available when Data Governance and Data Architecture are in alignment.
ENROLL IN OUR LIVE ONLINE DATA GOVERNANCE TRAINING
Join our three-day seminar to advance your Data Governance knowledge and become a CDMP specialist.
What is Data Architecture?
Burbank defined Data Architecture according to the DAMA International Data Management Body of Knowledge (DMBoK2):
“Data Architecture is fundamental to Data Management. Because most organizations have more data than individual people can comprehend, it is necessary to represent organizational data at different levels of abstraction so that it can be understood, and management can make decisions about it.”
An organization’s Data Architecture is described by an integrated collection of master design documents that govern how data is collected, stored, arranged, used, and moved. “So, even in the definition of architecture, we’re talking about governance, because there is a natural overlap,” she said. Governing information is very difficult without good architecture.
One of the respondents in a DATAVERSITY report entitled Trends in Data Architecture said that Data Architecture is “The science of assessing your data; but also the art,” and Burbank agrees. Governance professionals and data architects need to find a balance between business, which is more fluid, and architecture, which is more rigid, she said. Getting that balance right is indeed a bit of an art and a science.
What is Data Governance?
The DMBoK2 defines Data Governance as “the exercise of authority, control, and shared decision-making over the management of data assets.” Burbank describes the authority and control as “the stick” and the shared decision-making as “the carrot.” “It’s often governance that helps a project sing, and if you focus on the right business value, the right case studies and the right audience, it can really help things expand across the business.”
Data Architecture: Part of a Wider Data Strategy
Burbank discussed Global Data Strategy’s framework document for Data Architecture and Data Governance. At the very top is business strategy, which addresses alignment with business priorities and describes the reason for data use. “Are we trying to save lives with this data?” If so, it has to be very highly governed and regulated. “Or are we just doing some social media sentiment analysis, and let’s loosen up, and let’s not worry as much?” Looking at the framework from the bottom up, management is classified by what type of data it is (documents, unstructured, semi-structured, or unstructured data, etc.) and where it is located (in databases, big data platforms, etc.).
Business Drivers for Data Architecture Implementation
The Data Architecture report also found that regulatory compliance and governance were the leading drivers behind implementation of Data Architecture, along with business intelligence and reporting. In answer to a question about who is responsible for creating Data Architecture, as expected, over 90 percent of respondents said “the data architect,” but Burbank found it interesting to discover that enterprise and business architect, as well as data modeler and Data Governance Officer were also listed as responsible. Because data is becoming more of a business focus, more people are becoming involved in looking at the data, and collaboration is more important. “This sort of underscores that it takes a village to raise a Data Architecture.”
Find Balance in Implementing Data Architecture
Returning to the idea that Data Architecture is both an art and a science, she said if everything has to be fully developed and modeled before anything can move forward, nothing will get done. By the same token, if modeling or architecture gets bypassed in a rush to move forward, “It’s just the Wild West, and you end up spending longer. If you don’t have time to do it right, do you have time to do it again?”
Finding the right balance will provide business value, and where that balance sits depends on the industry. If it’s medical data, managing and modeling carefully is critical. If it’s social media data, there’s more flexibility.
Look for Business Value Levers to Identify Quick Wins
Identify and focus on areas where highest business value can be attained, she said, and provide support to a high-visibility product launch or initiative. “Read the company annual report, go to the CEO’s presentations and hear what they have to say, look at the corporate values and make sure that you’re really aligned with that.”
New initiatives that further company goals and values are more likely to catch the attention of potential champions within the ranks of company decision-makers as well. Building a model around key areas of business value can show areas where data can be used to provide quick wins.
Modern Organizations Use a Wide Variety of Data Platforms
A DATAVERSITY report entitled Trends in Data Management asked executives what data platforms they use, and although there is much discussion about the demise of the relational database, Burbank said it is still going strong, with over 80 percent of respondents reporting that they continue to use relational on-premise database technology.
Another finding made her cringe, she said: over 70 percent still use spreadsheets. “I don’t know if we’ll ever fix that. They have their place.”
Implement “Just Enough” Data Governance
Burbank showed a diagram of a pyramid, illustrating how not all data should be governed the same way. Flexibility is key. At the top of the pyramid is the smallest set, which includes master data, reference data, a common list of product codes across the company that everyone uses — anything that should be highly governed and mastered. On the bottom of the pyramid is exploratory data, sandbox, raw, and external data.
“Don’t crack down so hard on sandboxes that people can’t do anything, but then don’t under-govern your master data so no one knows even how many customers you have,” she said. Having governed master data available for use can help people working at lower levels doing exploration as well.
Crowdsourcing Governance and Metadata Definitions
Burbank explains differing levels of Master Data Governance as “the encyclopedia” versus “Wikipedia.” Some definitions have to be vetted, clearly and universally understood, and sent through a steering committee for governance: “This is the definition of ‘customer,’ and these are the six fields that are mastered — don’t touch them.” Other data is better for self-service and sandbox data use. It’s important to be flexible enough to meet the needs of different users.
Glossaries, Metadata Catalogs, and the Self-Service User
Self-service users want a mix of choices, so offer solutions that provide what they need:
- Corporate documentation
- Business glossary with common definitions
- Standardized list of what types of data are available
- Tagging, chat, collaboration
Build Data and Process Models
Use leading questions to create conceptual and logical data models to tell a story visually. “That’s a great way to get the story out to business users,” some of whom are intimidated by boxes and lines, so add relevant pictures. To the data modeler, it still looks like a model, but it’s a friendlier approach for nontechnical people.
Models can be used to clear up confusion in situations where a single common process is called by multiple different names, for example. Process models are also useful to clarify the customer buying experience or to understand where and how data is created and used throughout the organization.
CRUD Matrix: Create, Read, Update, Delete
The process of listing every place where data is created, read, updated, and deleted can provide insight and surprises. A CRUD matrix can show, for example, how an address change is entered but subsequently overwritten by someone else, or how and where multiple people are creating the same data. This process can also uncover critical data entered manually and stored on spreadsheets. Don’t forget about manual data, she said, “Because that is often the stuff people are using.”
Optimizing Restaurant Revenue by Linking Data with Business Processes
She used models and data mapping to help a restaurant optimize its menu. Their core business identity was focused on having a dynamic, creative menu, but they struggled with consistent naming from the ingredients list, to the POS system, and all the way to the printed menu.
She used business process diagrams to map out the flow of information, and CRUD matrixes to see where data was used, owned, and managed. They discovered that the chef had his own “data model,” with all of his menu ingredients listed on a whiteboard. Their supply chain had the same ingredients, but under different names. By the time information got on the menu, names had changed again.
The solution included a single view of the menu, consolidated and managed as master data in a central hub, with governance in place to outline workflow and policies.
She pointed out that the chef’s understanding of data may be surprising to some. “He had what I would have called the ‘data model’ of his ingredients and products, and how it all fit together. So don’t underestimate business people as not being ‘data people.’ They live data every day.”
Data Governance through Data Architecture
She assisted a retail vendor wanting to enhance customer experience, use IoT product data to improve product design and customer service, and optimize their product supply chain and delivery. The process uncovered several things that could have been fixed with some simple models and collaboration, she said:
- Key systems didn’t talk to one another
- Systems forgotten due to siloing
- Inconsistent data models in use
- Product codes changed and bypassed governance, shutting down the sales system.
Using agile methodology within a four-week timeline, Burbank created a process model, a customer journey map, a data model, a data flow diagram, and an organizational diagram. The company realized a structure and plan to reach their goal, and illustrated the need for master Data Management, Data Governance, and Data Architecture.
The process was a rewarding one. The Chief Marketing Officer told Burbank that, as a business person, she never expected to use terms like “data flow diagram,” but the model was something she could use and understand. The head of the sales department was able to see how the consistency of the customer information his sales people entered impacted customer experience. “He asked, ‘Shouldn’t I be governing how my guys enter this data? Because we’re going to use that downstream.’”
The Critical Role of Data Architecture and Governance
“When everything flows nicely together,” she said, a business data model can provide answers:
- What do I mean by “customer”?
- What do I mean by a “claim”? Is it a marketing claim? Is it an insurance claim?
- Do I understand the process that it’s a part of?
- Do I understand where data is used?
- What are my business rules and policies?
- Where is PII? What is PII? Is it this field and this database?
- What are the fields we care about?
Burbank recommends starting with a story, then building out the Data Architecture while collaborating with the right people to do some whiteboard design thinking. “And be flexible!”
Want to learn more about DATAVERSITY’s upcoming events? Check out our current lineup of online and face-to-face conferences here.
Here is the video of the Data Architecture Online Presentation:
Image used under license from Shutterstock.com