Data governance provides the framework for how data is used, who makes decisions about how it’s used, and who manages it. And as artificial intelligence (AI) and machine learning (ML) further infiltrate organizations, it has become all the more pressing for organizations to get a better handle on the vast quantities of often unstructured data generated by their customers’ activities. Done right, an effective data governance program leads to what all organizations are ultimately after: a single, consistent view of customers and their journeys.
Data is the “fuel” that supports business issues, said Scott Taylor, aka “The Data Whisperer” and founder of MetaMeta Consulting, which advises clients on master data initiatives. Business needs are often the catalyst to implement data governance within an organization.
“It’s an enabler,” he said. “Companies are trying to do one of three things: grow, improve, or protect their business. Data can do all three of those … and it allows you to scale.”
AIG is a good example. When he was head of data strategy and governance operationalization at the global insurance giant, Peter Kapur established a data governance program based on the need for GDPR compliance and data privacy. Once the program was implemented, it had an effect on other parts of the organization.
“We were able to reuse and leverage data assets created during that exercise for the rest of the company, [including] applications, inventory, and business process data flows by line of business,” said Kapur, now founder and managing partner of True North Data Management, a consulting firm that helps companies operationalize data.
Effective data governance encompasses people, process, data, and technology, in that order, Kapur explained. He said companies need to nail down the people and process part of the equation to foster adoption of any data governance strategy.
“[The] CIO is a critical part [of that strategy],” Kapur said. “The CIO is the custodian of the data technologies.” However, he cautioned that effective governance initiatives “need to be business-centric. It must be about the business.”
That’s why, according to experts, data governance needs to be a partnership among several senior business leaders, including the CIO, chief data officer, or chief information security officer.
“Regardless of where it launches, it needs to be enterprisewide and take into account all stakeholders, particularly the producers and consumers of data,” he told CMO.com. “For example, predictive analytics requires multiple data points from across the organization. The only way to do that is to pull people together through a data governance framework. The enterprise and lines of business [ideally would] operate as true partners.”
Stefan Zutt, head of information and communications technology and CIO at Green Climate Fund, an organization dedicated to helping developing countries reduce greenhouse gas emissions, agreed. Data governance requires support throughout the organization, including a “house-wide data classification policy and its underpinning procedures and tools,” he said.
Communication And Consistency Are Key
Equally critical is communication, added Michael Pascullo, staff VP, business systems at Atlas Air, which specializes in freight and passenger charters. Companies must “set guidelines and rules and communicate them,” he said.
Data availability, usability, integrity, and security of data in an enterprise are all crucial elements to any data governance initiative. Consistency, a defined way to capture information, and accepted values and ranges for data quality are paramount, as well. In support of these, many organizations set up a data governance council to support constituencies across the business.
“It’s important that the right people have access to data when they need it,” Pascullo told CMO.com. Appropriate access is the goal: not too much, not too little. Companies also want to assure data accuracy, he added, and it should be validated. “It has to be auditable,” Pascullo said. “You need to make sure the data is clear and it’s defined.”
In addition to the CIO, other IT stakeholders involved in data governance might include security and privacy professionals, data scientists, and analysts. Line-of-business leaders also participate, depending on the nature of the business driver. If it’s customer analytics, for example, marketing, sales, finance, legal, and customer service are a likely part of the council or advisory group, according to experts.
“The work needs to be coordinated with the CISO and the legal counsel and, if applicable, a risk management team,” regarding the parts of the initiative that deal with data classification, taxonomy, compliance and information security [CIA], and privacy, Zutt said.
Data-Driven Operating Model
Mark Picone, vice president, information and data services at Adobe, has spent the past two years putting the company’s data governance program in place. (CMO.com is owned by Adobe.) He agreed that both business stewards, who own the data, and technical stewards, who produce and manage the data, are equally necessary to successful data governance.
Picone and his team created a data-driven operating model (DDOM), in which all business lines are consistent in how they measure outcomes and handle data governance.
“We created a single view of the customer, at the most granular level, at a scale that had never been done at Adobe,” he said, adding that DDOM covers every geography, route to market, offering, and product across the company’s $7 billion global digital media business.
Adobe collects its customer data from a wide variety of sources, including anonymous Web browsing, email opens and click-throughs, product downloads, trial and paid subscriptions, and petabytes of product usage data. “We wanted to collapse siloed databases across multiple organizations into a consistent and governed view across the entire customer journey,” Picone told CMO.com.
Clearly understanding its business use case and the data required was the first step. Picone created a set of KPIs across five consumer journey stages. “Having a focus around the customer journey stages allowed us to to not boil the ocean, since our focus was so broad the customer journey stages created discrete, value-based business outcomes,” he said. “This adaptive form of data governance gave us guardrails to align our deliverables with business expectations.”
The process was challenging. It was important to gain the business’ sponsorship, as well as participation across finance, marketing, sales, customer support, and individual business units. Change management was also a big issue. “Those things are hard in the beginning, but once you have the assets aligned with critical mass, you create a form of data gravity whereby business and technical users see the value and adopt naturally,” Picone said. “The data governance process then takes on a life of its own driven by business users who gain value from it.”
That’s exactly what happened: DDOM fundamentally changed how Adobe runs its digital media business. “We have a set of analytical applications all leveraging a single source of truth across the customer journey that is used by hundreds of business users across the enterprise,” Picone said. “Further, we have operationalized the use of this data with defined processes and actions around each journey step all aimed at business outcomes.”
The Way Forward
A data governance road map ultimately affords companies a holistic view of their data. Data governance can be a valuable asset that contributes to revenue growth, cost savings, and operating efficiencies, especially as technologies like AI proliferate.
“If behavior or patterns detected are wrong, it’ll adjust incorrectly. It’s thinking for you based on data, so the data has to be right,” Atlas Air’s Pascullo explained. Data governance “sets the foundation for where you want to go.”
The best senior IT executives, he said, will strike a balance between satisfying business demands and maintaining data integrity. It’s important that the right people have access to data when they need it.
Measuring that success is “having zero instances where incorrect data is being reported in any capacity,” Pascullo said. “That’s success—not having any defects with any data reporting. [So is] having happy customers who are getting what they need in a timely fashion is another measure.”