Gavin Day, Senior Director of Data Management at SAS, discussed the increasing importance of data quality and governance and how master data management can solve a number of problems.
Argyle Executive Forum: Let’s start out with some background. Tell us a little bit about SAS and your role as Senior Director of Data Management.
Gavin Day: SAS delivers advanced analytics that turns data about customers, performance, financials and more into meaningful information. The bottom line for SAS is that we transform the way our customers do business and make decisions. I believe we’ve been successful doing so because of our commitment to customers. We involve our customers throughout the R&D process to ensure it’s meeting their needs.
In my role at SAS, I’m responsible for the SAS go-to-market strategy on data management. I work very closely with the unbelievably smart teams in R&D and product management to make sure our technology is answering the needs of the market. I work with customers across the Americas to help educate and provide advice on topics like data management, data governance and master data management.
It looks like you enjoyed a long career in IT at DataFlux; what eventually attracted you to the role you fill today with SAS?
Technology is absolutely my passion and has been for a very long time. I was on the IT side of the house early in my career and learned a tremendous amount working with my internal customers as well as working with CIOs from all over the world. Over the years, my roles began to evolve to include responsibility for professional services, technical support and sales. No matter where I went and what project I worked on, eventually all roads pointed back at the data and information within our organizations. All too often, the topic of data management becomes an IT vs. the business discussion, instead of the collaborative approach that most of us know we need but is culturally hard to make successful. My experience on both the IT and business side of the house has helped me gain an interesting perspective. There’s no question that if we had to walk in the shoes of our business or IT counterparts we’d have a better view of our organizational needs.
“As CIOs and CTOs struggle with all the “V’s” of Big Data (volume, variety, velocity), they need to recognize that the business struggle is all about data analysis.”
One theme that is addressed at our CIO Leadership Forums across the nation is how to effectively manage big data. What should IT executives be cognizant of as they decipher the best way of managing big data in a manner that adds the most value to the business?
As CIOs and CTOs struggle with all the “V’s” of big data (volume, variety, velocity), they need to recognize that the business struggle is all about data analysis. Much of the basic blocking and tackling of big data deals with processing and storing data. The key to transforming that data into meaningful analysis is data management expertise. For example, managing data for predictive analytics has unique requirements not present for historical data analysis.
Could you give us an overview of why data quality and data governance are increasingly important issues for CIOs to address?
The primary driver for CIOs is that there is a fundamental lack of trust with data, especially in areas of the business that haven’t been forced to respond to compliance mandates.
The secondary driver typically is that the business has considered data as being an IT asset. This was due to the way that data has historically been tied to systems, applications and data warehouses that were primarily managed, built and provided by IT. As data demands have increased, the mechanisms to store and interact with data have changed. Data management has matured and data is absent of the historical constraints, and as a result the business expects that IT will continue to manage and govern data. CIOs understand that without a business partnership they simply cannot succeed. The rationale is that without the ability to manage data in a business context, the data cannot be controlled in a way that will be accepted by the business, and data proliferation outside of IT management and lack of trust will continue. Data governance provides the mechanism for IT to adhere to guidelines dictated by the business, and as a result, the data meets business expectations. Data governance also provides the vehicle to manage data change in an effective manner. You often hear data governance and data quality used together as it relates to data. The business sees the data as having issues and not meeting its needs, so it assumes that the quality of the data is poor.
This could certainly be due to the fact that the proper data quality controls and technologies haven’t been applied to the data, and the old phrase “garbage in and garbage out” begins to have applicability. On the other hand, the words “data quality” can be used due to the lack of business context. For example, dimensionally (completeness, conformity, consistency, accuracy, duplication, integrity) the data could be of high quality but low in business context. For example, if I receive a bill for services that is understated by $20 for parts that were purchased in addition to the services rendered, one might state that the bill isn’t accurate.
How has the complexity of data changed (if at all) over the years and how is this affecting the role and responsibilities of the CIO?
Data is more complex in terms of volume and the number of variables that decisions can be based on. For example, in the past, marketing campaign segments would be based on gender, age and household income. Today, there are dozens of customer attributes along with a large volume of customer behavior data that can be stored and analyzed. This leads to an exponential rise in the number of possible segments and marketing strategies. Add to this the many marketing channels and product mixes, and you have a very complex set of variables to analyze. The role of the CIO is to make this complex data easily accessible, and also to provide business users with the ability to easily analyze and visualize the patterns in the data and bring meaning to it. The role of the CIO used to be making sure each line of business has the hardware and applications required to run its line of business. Today, a CIO must have a higher level of thinking. A CIO must visualize the business as an entity consisting of a multitude of real-time data streams. Hardware and software become not the objectives, but a means to manage these constant data streams. Without a clear understanding of the data coming and going and its value to the business, decisions might be made that suboptimize the use and value of the data flowing across the organization.
As technology is being integrated across the enterprise, the volume of data generated has grown exponentially, and access to this data is more important than ever. What are some of the major obstacles currently hindering access to data for business leaders across the enterprise?
We often hear that organizations are data rich and information poor; this is contrary to what one would expect from today’s leaders. Most organizations would say that they have more data than ever and less information. You hear this for a variety of reasons but mostly due to the lack of enterprise data management. Most organizations can manage data effectively to support a department or a line of business, but few can effectively manage enterprise data. Why is that? There are conflicts with data definitions, ownership, applicability, metadata, levels of data maturity and lineage across the organization. The next challenge is finding the business benefit of enterprise data. No one doubts the value, but when the rubber meets the road and all parties are in agreement, the data is so watered down that the business value suffers. The trick will be to strike a balance where the data isn’t a data warehouse dumping ground, but is agile enough to meet business needs without being out of date when it’s needed.
Metadata – clear business definitions around meaning, usage and content
Data Lineage – where did the data originate from and what if anything has changed it
Different versions of the same business entity (e.g. product, customer, vendor, location)
Data Architecture – fragmented data architecture creates integration challenges and increased costs
How do solutions like master data management (MDM) help the IT department alleviate this problem?
Solutions like master data management allow organizations to put their best foot forward in the management of data. MDM brings active data management to the business and to IT. The traditional IT focus on hardware and software functionality and not on data functionality created the situation where the data was not being managed for enterprise use. Data management was more about managing the transactional systems, whereas MDM is about active management of the data that represents the business entities. MDM requires enterprisewide support in defining the business entity metadata, in understanding where the data comes from and how it will be used after the MDM process. MDM forces IT and business to work together to understand the data streams and to create a plan on how to integrate and use the shared master data.
“As data demands have increased, the mechanisms to store and interact with data have changed. Data management has matured and data is absent of the historical constraints, and as a result the business expects that IT will continue to manage and govern data.”
Why have companies struggled with successfully integrating MDM into their infrastructure, and what are some solutions to this problem?
Companies struggle with MDM because systems and data are fragmented across the organization. There are different business owners of the data, disparate systems and technologies, different meanings for what seems to be the same data, and different storage architectures that create obstacles.
Another issue is the problem of data quality. Each data owner, each transactional system has different business rules and levels of data quality. When you try to integrate the data, these differences in data quality suddenly become apparent, whereas before they were not so important. These islands of data create data quality problems that negatively affect the enterprise’s ability to create and maintain master data that can be published back to these same originating systems.
What are some of the major benefits of MDM solutions when it comes to cost, efficiency and overall agility of the organization?
The value of MDM is in the ability to effectively make decisions about customers and products. MDM forces active management of the enterprise data so there is one consistent view of the key business entities. MDM, by providing a single trusted view, increases the ability to sell to the customer, speeds market decisions and forces inefficiencies in business processes to improve. Customers feel they are better understood because you are touching them based on an accurate picture of the customer and the products they have bought or shown an interest in.
The organization is more agile because there is a common understanding (common metadata) about the key business entities, so strategic decisions can be made quicker. More time is devoted to discussing and making decisions and less about learning what the data means, figuring out if it is correct or where it came from.
MDM increases the data quality across the enterprise, so risk, compliance costs and resources are reduced. Data quality costs will decrease over time as MDM requires process changes that remove the need for manual audits and data corrections.
MDM benefits analytics and reporting; instead of integrating bad quality data on the analytics platform, the analytical resources can be focused on providing the business users with a wide variety of data that can be easily analyzed, creating opportunities for revenue growth.
For more details about data management solutions: sas.com/software/data-management
Get fresh perspectives on data management from experts writing on the Data Roundtable blog: blogs.sas.com/content/datamanagement/
Gavin Day is responsible for SAS Americas sales and operations for the Data Management driver team, which includes sales specialists, systems engineers, solutions architects and a Management Consulting Services group. He also works closely with the SAS Research and Development group to deliver customer-driven solutions. He joined DataFlux in 1999, and SAS in 2011. He has helped the company design, implement and maintain enterprise-level data management solutions for companies worldwide.
Visit Argyle Executive Forum's 2019 CIO Think Tank: The IT Revolution in Healthcare in New York, NY on Jan 23, 2019