5 lessons for organizations that use Power BI
The last two and a half years I have been helping organizations adopt and implement Power BI. With the right approach Power BI can be a huge success. But if you are not sure how to do it right, you might make the data landscape messier and more confusing.
Here are the 5 simple lessons I have learned:
- Make sure to distinct the valid 'one version of the truth' data from the raw data. This distinction must be clear for your users. They need to know what data they can play with, and what data cannot be played with. Analysis services cubes/tabular models are a great platform to service the enterprise validated data, whereas a data lake or SQL databases can be a great source for users to make their own models on.
- Make sure the users can find the data. Catalog your data sets. Azure data catalog can be interesting, but whatever tool you are going to use, it is important that everyone knows where to find what data sets he or she can use and who owns that data.
- For users who want to do self service BI: organize an on boarding training so that users know how to avoid mistakes. If they do not understand the tool they might make wrong data relations and draw the wrong conclusions about their data. This can lead to wrong decisions.
- Power BI is not your data warehouse, nor will it replace it. Data quality and one version of the truth is not what you get out of Power BI. In the end, when you save in Power BI desktop, you get a report file. A report is not your enterprise data model.
- To help adoption, you need to have enthusiastic people from the business that want to create reports for their departments and colleagues. They can be ambassadors for a data driven organization. IT should not decide which reports need to be created. Create the reports with them, not as a customer but as a co-creator.