Data modelling in Power BI – basis for effective reporting
Unfortunately, too little attention is paid to data modelling in many Business Intelligence (BI) projects and its value is underestimated. Above all, one wants to show results quickly – i.e. meaningful reports with diagrams and tables. However, this can quickly take its toll, as reports based on a poor data model can usually only be expanded or managed with great effort. New BI tools such as Power BI do not solve the entire problem, but they do support the data modelling process with innovative functionalities.
Who hasn’t experienced this? You start a BI project with a catalogue of requirements that is not always complete. But the clients don’t want to wait so long until everything is defined, but for various reasons want to see results as soon as possible. The team gets to work and needs (too) much time for the ETL phase (Extract, Transform, Load), because the interfaces to the various data sources are not so simple after all or were not documented enough.
Although enough time had been planned for the data modelling phase, since ETL took longer, the data modelling had to be “shortened” in order to create time for the visualisation. The client sees the reports, but not the data model. In the end, the client is satisfied, but already wants extensions that should be implemented “tomorrow”. And then the problem starts all over again. The data model works, but any extensions make the data model susceptible to inconsistencies or have to be programmed laboriously. The effort increases disproportionately and reporting becomes error-prone.
Microsoft Power BI, as a third-generation BI tool, does not solve the whole problem, but supports the process of data modelling comprehensively through innovative functionality. From the automatic recognition of relationships to the visual support of the Model Viewer and the optimisation and verification of the data model – a toolbox and an active and excellent community are available at all times, making data modelling easier and faster.
Especially in the autumn updates of Power BI, data modelling has been greatly expanded and enhanced. The new Model View makes it easy to navigate and quickly identify dependencies. The good practice collection from the community is an invaluable resource, and I personally have already received many suggestions on how to improve data modelling.
f you want to learn more about Power BI and if you also want to put your data modelling on a new basis, then contact us. We would be happy to talk about our experiences and support you in your reporting challenges.
Text by Johann Kurz, Business Development