Jason Dye, Director of Enterprise Data Governance at Ally, says that sometimes Data Governance feels like a thankless job. “I’m struggling up this sandy godforsaken hill and then some guy’s standing on the sidelines going, ‘I don’t know what’s so hard about Data Governance. It’s easy.’”
But that may be because Dye makes it look easy by building best practices into every step of the Data Governance program. He shared those best practices for meeting challenges to Data Governance with attendees of the DATAVERSITY® Enterprise Data World Conference, during his presentation titled Best Practices to Start and Maintain a Data Governance Program.
Challenges to Data Governance
Almost every company has disparate data systems. Conflicting data flows and a lack of data ownership can lead to a lack of trust in information, he said, and an inconsistent understanding of that information. According to Dye, challenges come from a variety of sources:
- Limited funding and resources, or competition for them
- Leaders who don’t buy-in
- Complicated, detailed and high-level dataflows
- Third party- and other data the company doesn’t control
- Partial access to data
- Fragmented or siloed IT and ops teams
- Incorrect preconceptions of what Data Governance is
- No tools — or weak tools
- Lack of knowledge by staff, internal or external to Data Governance team
Dye believes it’s possible to meet these challenges, and in the process, get funding, support, and recognition for the value of a governance program.
Meeting the Challenges
Dye presented a general overview of best practices, which include taking a long-term approach to meeting Data Governance challenges, working with human nature to change culture, ensuring that business needs are met, and treating data like a valued asset. Ensure that programs are sustainable and efficient by solving the root cause of problems and using automated tools to make processes more cost-effective. “Data Quality is the result of good Data Governance,” he said.
Meet Company Goals
He suggested starting by understanding the mindset of senior leaders and what they want, which often boils down to “better data for less money.” In order to deliver, increase the value of the data by:
- Creating trusted data source(s), with data that people can use
- Removing guesswork by being consistent and transparent
- Lowering costs through reduced reporting efforts (internal and regulatory), streamlined data flows, reduced project implementation times, and standardized data approaches, including systems, policies, procedures, and standards
- Providing a release valve by solving data problems at the source, and improving the monitoring and tracking of Data Quality and other data-related efforts
- Fulfilling regulatory and audit needs by providing proof of controls with documented policies and standards as well as correct lineage
- Automating processes to lessen risk and save time
Reach Team Goals
- Create Value in Data Governance: “The value of Data Governance isn’t, ‘Oh hey, we found some issues and now let’s go fix the data.’ It’s about finding what’s causing the problems, so the data isn’t wrong to begin with.” Creating self-sustaining Data Quality environments and solving underlying problems makes it possible to drive value and eliminate the process of “fighting an uphill battle with senior management all the time to get funding, to get new tools, to get your group in front of other areas,” he said.
- Change the data culture of the company: “This is a hard one,” he said, but as the message of Governance value is communicated over time, culture will change. Newsletters or other communication channels should communicate topics about existing and desired data culture, as well as highlighting areas where governance has helped reduce risk, save time, or perform better.
- Create team credibility across the company: Make sure your front-facing team members have good people and communication skills so the Data Governance area is a comfortable place to bring problems for everyone across the company. “You want those people to come back to the table later” and tell others that they want the Data Governance group involved “because they will help you solve problems.”
- Data Governance team goals and company goals should be separate, but interrelated: Team goals have to align with company goals in order to create value; “You’ve got to be pretty much in lock step.”Identify your mission for the program and post it on the wall.
Data Governance Tactics
It’s necessary to identify your Data Governance champion he said. Identify a senior leader as a Data Governance champion. A senior leader can provide “muscle,” and a way to get involvement from key personnel on the business side, and motivate reluctant team members. Talk with the business side and let them point to where problems exist, but “don’t try to boil the ocean,” he said.
To accomplish this he recommended starting with a pilot area that has meaning for your champion. Identify critical elements for the pilot area and assign owners for each element. Produce detailed lineage for each critical element, starting with those that have problems and will have the biggest potential impact.
It’s then necessary to create a senior level steering committee to establish policies, standards, and procedures around data, he said. Document data flows in that area at a high level to provide an understanding of data sources and where that data is stored. Create a Data Governance working group from vested parties.
And the to move onto solving the root cause of critical problems.Use a methodical testing process to solve problems, starting with the root cause of errors. “This is where Data Governance proves its worth and creates value,” Dye said. There may be a thousand issues, yet by seeing a common root cause for 20 or 200 and addressing that cause, groups of errors start falling away and the process becomes less daunting. Implement a communication strategy designed to build user confidence in the process.
Once that is successful then show the pilot’s success to prove value of program expansion. Document time saved to support effectiveness and efficiency of governance efforts, and translate the hours saved into actual dollars for better impact. “Rinse and repeat for the next business line or tackle more data issues in the same business line.”
Data Governance Tools
- Glossary: Should have a few thousand terms, easily accessed so all different business lines can use it
- Metadata Repository: Dye considers a repository extremely important, so that IT and business lines can “understand what they’re looking at.”
- Lineage: Lineage is essential for Data Quality testing at the table and column level. Present it in a graphical form as well as in a detailed repository.
- Data Quality rule writer: Serves as a reconciliation tool as well as a way to write Data Quality rules
- Reporting tool: A reporting tool with an issue tracker, reports, and metrics will allow different groups doing root cause analysis to quickly and easily share and monitor their findings, he said. “Things you want to take to senior management, things that you want to be able to show trend lines and Data Quality issues.”
Communication: A Constant Drumbeat
For best results, Dye recommended constant communication about Data Governance using four different levels of communication, based on role:
- Senior management gets high level metrics. If they are given too much information, he said, either they ignore it completely because they don’t have time to look at it, or they read it great detail and have a hundred questions. “So I try to give them what I think they need at a high enough level to make it impactful.”
- Mid-level management gets the same information as senior management but with more detail.
- Business/IT stewards and owners get detailed Data Quality issues, errors by record and by element, and trending of record and element errors, with six- and twelve-month views.
- Data Governance analysts get detailed Data Quality issues, errors by record and by element, and trending of record and element errors, as well as errors by system, by data flow, and by operations process.“The general idea here is that you want this detailed enough so they can use it for root cause analysis.”
Data Governance Team Structure
Dye said that all these roles will eventually be needed, but not all organizations will be able to have each as a separate position in the beginning:
- Technical position: responsible for understanding Data Quality rules, reconciliations, and lineage. Dye suggests filling this position with a detail-focused person who can quickly spot misspellings and inconsistencies.
- Business-facing position: responsible for identification of critical elements, definitions, glossary, and root cause analysis feedback loop.
- Communication position: responsible for metrics and reporting to senior leadership, mid-level managers, IT, business and Data Governance staff.
- Project manager: responsible for organization of each project and initiative.
- Leadership: responsible for providing vision and strategy, marketing, and driving the project.
Make Governance a Trusted Partner
Ultimately, he suggests working with human nature to engender cooperation and change the perception of the Governance process. Data Governance is often seen as an expensive, time-consuming add-on to a project or program, rather than something that offers value and is built into the planning process. To mitigate this, he said, his team shows up and offers all the existing lineage relevant to the project that is currently available: definitions, owners, locations, trusted sources, and where they’re testing, down to the table and column level of where relevant data is located. Because updating lineage or adding terms to an existing glossary is much easier than finding all the necessary information and building a structure for it from scratch, Governance is perceived as less of a burden. “You don’t have to recreate the wheel. Use this, and that should help alleviate some of the cost and effort in your project,” he said.
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 Enterprise Data World Presentation:
Image used under license from Shutterstock.com