Enterprises are facing issues with the information represented in report. These range from incorrect consolidated figures, past data that is not relevant to using different names for the same category. As a result, there is a loss in trust and therefore a reduction in the use of business intelligence applications. In this situation, the enterprise is then at risk for driving its activity.
Even if an enterprise datawarehouse is enabled to homogenize information and to reduce arguments about reports accuracy, it must be stated that search for coherence can be improved and boosted thanks to analytical MDM.
Analytical MDM benefits
Analytical MDM aims at moving the management of Master Data from datawarehouse’s technical processes to a business-oriented application with governance processes. Hence, analytical MDM enables to provide core elements to business intelligence applications: which helps on the dimensions of quality, cross-references and hierarchies.
In a traditional approach without analytical MDM, the evolution of business rules around dimensions requires to go back to a heavy development cycle with IT teams. The analytical MDM enables to involve the business users in the dynamics of build, evolution and use of the business intelligence applications.
Analysis axis (dimensions) of quality
Intrinsic values of dimension attributes must be of quality. There must not be duplicate members in the dimensions. The structure must be coherent (conformed dimensions).
It is to highlight that analytical MDM is also a lever for enabling predictive analysis in the sense that business users may enter effective dates in the future while administrating the dimensions to forecast effectively.
Cross-references are required to link figures (facts) to unique labels (dimensions). This is particularly useful when the datawarehouse has different source applications.
For example, ‘USA’ is the best label that we want to see across all the reports. This can be achieved with the following cross-reference: ‘USA’ in application 1 corresponds to ‘US’ in application 2.
Consolidating the information and navigating through it may be done in different ways. Let’s take the example of the “Geography” dimension: the sales director wants consolidated figures by territories and by territory groups whereas the central administration asks by town, county, and state. Two hierarchies must then be defined.
Architecture: the MDM brick for the datawarehouse
We see in this architecture that analytical MDM is not intrusive with regards to the operational applications (sources). The analytical MDM consolidates the Master Data issued from the operational applications. Analytical MDM usually relies on a normalized datamodel whereas dimensions are denormalized by nature. A denormalization operation must be considered when providing the dimensions to the datawarehouse.
Analytical MDM, a first step to enterprise MDM?
The answer is “Yes, certainly”. In many companies, the datawarehouse is the only location which tempts to offer a unified view of the information. As sawn earlier, analytical MDM is a brick which helps on achieving a unified view and with quality.
Consolidated Master Data which is a result of the implementation of analytical MDM are then true golden references for the whole company. By synchronizing operational applications with the MDM, the latter becomes also an “operational” MDM. An enterprise MDM (operational + analytical) enables to maintain the coherence of the information through: the operational information system, the analytical information system, and between both of them. The two worlds are not disconnected anymore. Starting with analytical then expanding to an enterprise MDM is a good tactical approach which lowers project risks and secures users’ adoption.