Data governance challenges arise in almost every organization. Data has the potential to be your biggest competitive advantage, so it makes sense to invest in analytics, machine learning, artificial intelligence and other ways of extracting insights from the data an organization produces. But if you don’t have a good data governance framework in place and haven’t emphasized that governance is everybody’s business, then the various challenges that come with data governance can cripple that competitive advantage. To use data as a source of intelligence and a strategic advantage, your data must be easily understood and consumed. But at the same time, you need guardrails in place so that people use data in ways that don’t put the organization at risk.
1. Understanding the business value of data governance
Organizations are constantly generating data. The sheer volume of data in modern organizations requires digital transformation initiatives to manipulate and serve up that data. Understanding the business value is especially challenging in older, established organizations that are adopting digital transformation more slowly. However, an advantage of implementing digital transformation is the ability to share relevant data with entities like employees, customers and vendors. This data democratization holds potential for moving the business forward, but it relies heavily on having the guardrails of data governance in place.
One solution has been the emergence of the chief data officer (CDO). This role started as a data-focused, right-hand person working for the chief information officer (CIO). With time, the CDO has risen to peer status with the CIO, and is tasked with making sure that the organization’s thirst for data is quenched efficiently. That entails rolling out an architecture and a data platform that enforces data governance and keeps it top-of-mind.
The analytics group can also reinforce the business value of data governance. Often, this group is affected the most by low-quality, misunderstood, or misused data, as they are frequently charged with delivering insights, upon which crucial decisions are made.
2. People think IT owns the data
In the early days, many organizations relegated data governance to IT, thinking it was a matter of putting rules in place and restricting access to data. But data governance was rarely prioritized, and didn’t survive IT budget cuts because it was not treated as having real business value.Much of data governance is an effort to clarify who really owns data. The idea that IT owns data — and, therefore, data governance — is an obstacle to digital transformation. In any decision about data, the first consideration is the business perspective, with IT enabling that decision afterwards. To better illustrate, consider the role of the database administrator: DBAs own the database, but they don’t own the data itself. Their remit is limited to administering the vehicle that holds the data. IT should not be making decisions about implementing solutions that affect data without talking to the business about the impact of those solutions. That would be going backwards, in an unsustainable process. So, to achieve success, smart organizations have moved data governance back into the business, started showing stakeholders the risk of not having it, and started demonstrating value it can bring to an organization. They’ve also started defining data ownership, which sits with the managers in the business organization. A large utility in the European Union successfully made this shift, and realized savings of 30% on external data management costs, 50% reduction in data discovery time and 8 million Euros in business impact saved in first 18 months. It’s not a matter of who owns Oracle data or who owns SQL Server data; it’s a matter of sales owning sales data, marketing owning marketing data and so forth.To overcome this data governance challenge, a best practice would be to appoint a data steward who works with a DBA to provide access to that data. That ensures that you’re neither giving people random access to data nor fighting logjams with DBAs.
3. Limited or misallocated resources
Once the C-club members see that data governance is important, the business is driving governance and IT is responding, the next data governance challenge becomes limited resources and misallocated resources. Whom will you get to take on the new roles of Data Owner and Data Steward? Ideally, when data governance is up to full speed and full scale, those roles are filled by dedicated employees. Designating a data owner isn’t as difficult because it means having a senior-level decision maker make a few more decisions. But finding a data steward is trickier. The shortage of skills is a big obstacle to achieving ROI in data governance, so most organizations try to train somebody with tribal knowledge to be data steward. As most organizations start out, the role of data steward is a part-time job staffed by somebody on the business side who frequently works with data. On the IT side, it’s usually a data architect or data analyst with additional, part-time responsibilities. Business analysts and business intelligence (BI) specialists are good candidates for data stewards because they’re familiar with the data and with the technical folks on the team. Your roadmap and plan should ensure that, once you’re demonstrating the ROI of data governance, you start to dedicate people to these roles. Large organizations further down the road may have eight to 10 dedicated data stewards who cross the lines of business. For organizations without the resources to have full time data stewards, one solution is to use consultants; not IT consultants or outsourced developers, but data governance professionals from specialty firms. Some consulting firms are building practices around data governance experts who can fill those roles. At a small scale and for a limited period of time, it’s a great way to achieve early success.
4. Siloed data
In many ways, siloed data happens because of different approaches to data operations and versions of technology. One of the things that most companies avoid whenever they modernize is figuring out how to migrate their systems holding legacy data. Those systems still do the job, and ripping and replacing them is prohibitive. Additionally, if you don’t know what’s that legacy data contains, then it’s less of an expense to try to solve the problem through downstream integration. For example, traditional relational databases replace a lot of what was on the mainframe. They speed up transaction processing but require highly structured data. So, what do you do with all your new, unstructured data like video and social media? You can put it into NoSQL databases, but that creates a new silo, so you turn instead to hybrid databases that offer the best of both worlds in one database. Data silos are a natural phenomenon. Think of them from a business perspective. Often, the business itself is siloed, and the people focused on transactions don’t communicate with those focused on strategy. You’ve already got silos around Marketing, Sales, Manufacturing, Engineering and so on, so it makes sense that their data would be siloed as well. Breaking out of those silos is key to further leveraging the data an organization produces.
One solution is to collect all of the information about the different types of data on different technologies for different uses by different parts of the business. With the right data governance tools, you can connect the metadata of different data types and see it in a uniform fashion. You can then map the architecture of your infrastructure and keep it updated more dynamically than if you tried to capture it in static documentation. You replace your documentation by collecting it from the database instance itself, then bring it into an environment where you can analyze it and make it useful. Metadata management is fundamental to successfully operationalizing data governance. Business users, of course, do not care whether the data resides on Oracle, SQL Server, Mongo DB or any other platform. They only want to understand how the data flows through their business and the point at which it comes to them.