Over the last year we’ve seen a major shift in the insurgence of business users performing their own data analysis to drive “informed” business decisions. This shift comes as a result of the ever-accelerating speed of business operations, and it’s here to stay. What was previously termed shadow BI departments in the early 2000’s, are now being reinvented as business user lead self-service analytics, answering the call of data-hungry business leaders who rely on analytics to manage performance.
However, many of the same issues that plagued departmental analysis before still exist today. Users are often meshing data from various sources without the guidance of internal IT, taking shortcuts, performing uncorrelated and incomplete analysis to produce imperfect results.
It has now become an expectation of these business users to solve major business challenges, predict market outcomes, and make transformative recommendations using self-service analysis, but without proper guidelines, governance or policies, the output of self-service analytics isn’t reliable.
Self-service was established to alleviate constant IT requests and empower business users with access to work with corporate data, even if they lacked the expertise in business intelligence, data mining, or statistical analysis. Companies have introduced self-service in areas of frequently used data, giving users access to customizable reports and at times even direct data access to iterate it and draw conclusions that support their specific business function. The removal of IT from the equation, however, has led to inconsistent findings, poor data quality, and worse, internal security breaches.
- Reliance on inaccurate or stale data
- Unusable data models
- Flawed business logic
- Inconsistent results across business units
- Failure to verify data
- Diminished credibility
- Sudden report errors
- Compliance penalties
- Data security threats
It’s not surprising that creating data assumptions and ad-hoc analysis outside the governance of IT can result in flawed analytics but even the IT department is supporting self-service, in theory, these days. At the surface, it sounds like a viable solution to scaling IT – and it can be if done correctly – but most IT departments are rushing to get reports and data marts stood up, knowing that there are inaccuracies in the data that aren’t being properly accounted for and knowing that the data will not be maintained.
Though self-service initially helps alleviate IT workloads, ultimately when these independently-maintained reports and dashboards break, it falls back on IT to fix them. In essence, these self-service reports have become a part of the larger BI program portfolio. BI teams often don’t account for this kind of support and maintenance, leaving business users feeling unsupported and producing flawed analytics.
Data governance is the management of the usability, integrity, availability, and security of data to ensure proper control throughout the enterprise. Governance is more than just a BI team’s responsibility, but rather, any department that is using data – which today is most of them. Whether HR is uploading confidential consumer and employee information or marketing is merging CRM and email data, unsecured data can create risks. Establishing a single source of the truth equates to creating policies and reference documentation for the use of all data and privacy measures.
Prior to establishing documentation, enterprises looking to incorporate policies for self-service should define the key metrics for determining success, privileges necessary for accessing private data, processes for creating and sharing reports, and maintenance of data quality and privacy. This information combined, can then live in a data glossary or dictionary, typically created within an organization’s database management system software – like Microsoft SQL Server.
A data glossary allows users to reference definitions and ensures users have a place to go to for answers to their data-related questions. Enterprise Metadata Management, or ‘data about data’ is a common term to use when exploring a data glossary. Metadata uses physical, logical, and conceptual data perspectives which, when used in a repository, can explain, locate, and describe the use of an information resource. Creating a data glossary or data dictionary is a useful way for users to reference not only where the data is coming from, but how to use said data in analysis. Defining topics and data points on even basic terms ensures that there is a common understanding across the enterprise. Further tracking data sources is a way to establish proper data quality and legitimize outcomes.
Empowering users with governed ways to analyze data can alleviate IT strain, create a single version of the truth, and create valid assumptions that could, in the end, reduce poor business decisions.