For a strategic BI initiative, it is critically important at the outset to understand the relationships between data quality, data governance and master data management.
While there is much theory around these subjects, experience has shown that in practice it is essential to first implement effective data quality management, which drives and supports data governance (people and process) on the one hand, and MDM (architecture and technology) on the other.
Data quality management company InfoBlueprint has noticed that many South African companies embarking on recent initiatives to update and rationalise their existing BI infrastructure, or to consolidate multiple disparate data marts scattered across the organisation, are recognising these projects as ideal opportunities to address data quality and data governance issues that have plagued previous attempts at BI. Organisations have stated that in the past there was too much reliance on technology components alone to "sort out the data problems".
While most modern technology underpinning BI is quite capable of delivering on its objectives, it is now a well-known fact that, particularly in large organisations that have grown and evolved over many years, there are often serious challenges with harmonising, correcting and consolidating operational data. Through an associated lack of effective governance, over long time periods, data in such an environment typically exhibits ambiguous (or missing) ownership and accountability, non-standardised storage structures, a high degree of inconsistency and duplication and, therefore, questionable reliability.
Companies that are experiencing these symptoms as a reality in their BI renewal projects are therefore shifting emphasis from technology to people and process issues that have a direct and ongoing effect on data quality. Broadly speaking, this means introducing data governance to ensure that data is treated as a shared corporate asset.
Unfortunately, however, the term data governance is wide open to subjective interpretation and abuse, and even with the best of intentions it is a difficult topic for many companies. For example, data destined for use in BI is sourced from across the organisation - with different people and divisions typically having completely different perceptions about data`s meaning, accuracy and quality. Data governance is thus applicable not only to BI, but to the entire enterprise, and to both "business" and "IT". Consequently, attempts to establish it as a top-down initiative are most often thwarted by ill-conceived approaches, unmanaged policy setting, and a high degree of misunderstanding and resistance because of the inherent politics involved with data exploitation and perceived data "ownership", or lack thereof.
In order for data governance to take root in a company, it is necessary to formalise data accountabilities and to get people to spend time, money and effort on data issues that have previously been left unaddressed. To get people to do this requires a combination of executive level policy setting and "law making", combined with ongoing measurement and monitoring of the newly formalised roles and accountabilities. This in turn requires that these efforts show quick returns and value on the resources spent on data governance. The ideal candidate to be a catalyst for action is a BI project, which becomes the incubator for data governance.
However, critical to the success of both BI and its data governance offshoot, is a proven increase in the reliability of data, or put simply, a measured improvement in data quality. This means that the BI project itself must be used to establish a formal approach to the management and resolution of data quality (DQ) issues encountered on the project. Together, the goals for BI and the structures, disciplines and artefacts developed for DQ in the project, form the foundations of data governance, ultimately across the entire organisation.
From an architectural perspective, it is also important to consider the role of master data management (MDM) technology, and its potential contribution to efficiencies in not only the BI environment, but to ERP and all other operational systems in the enterprise. MDM allows a shared resource, such as customer data, to be effectively, accurately and consistently shared across all systems, including BI. The success of an MDM technology implementation, however, is also directly dependent on the quality of master data, and therefore once again effective data quality management within a formal data governance framework is a prerequisite for MDM success.
But because MDM is by definition an enterprise-wide endeavour, at risk of encountering the same political threats as data governance, it too can benefit from a testing ground, and an early beneficiary to demonstrate value once again is a BI project. Once MDM has proven its value to BI and analytics, it is much easier to establish MDM within the more sensitive and technically challenging operational system environment.
In summary, for a strategic BI initiative, it is critically important at the outset to understand the relationships between data quality, data governance and master data management. While there is much theory around these subjects, experience has shown that in practice it is essential to first implement effective data quality management, which drives and supports data governance (people and process) on the one hand, and MDM (architecture and technology) on the other.
Together, these three elements combine to produce a practical and holistic solution to the data and BI challenges faced by almost every organisation today.
Share