As we come to the ends of our season with Weekend Work, there are some lessons to consider at this point. Success in Power BI comes from testing your models completely before you release them, including your measures; it is with that in mind that this week we look at creating random tables of data and the risks posed by your measures being used beyond their initial purpose.
Create random tables of data to validate your aggregations and ensure that it does when you need your DAX to perform. Random generation is no long term strategy and always carries the element of risk that it will be based on your assumptions about the data. Use it but be prepared to build multiple versions to test your work better.
One of the biggest problems with DAX is also its biggest strength, portability. DAX measures can be used and reused often outside their intended purpose; this can lead to strange results. This video highlights just one simple example; it shows that you need to be aware of what your measures do and make sure you test the outputs and document any footnotes. Ultimately a measure is not in itself wrong; more, the context in which it is used is incorrect.