Why do I need a Good Data Model?

How to achieve efficient data models in Power BI desktop, with what to do, what not to do - including data hoarding.

You’ve generated all of your data tables using the 'Get Data' function in Power BI Desktop from various data sources. When you click on ‘Save and Exit’, Power BI will automatically refresh your data model, and you will see the below window in Power BI Desktop.

In most cases, you don’t have anything to worry about; however, if you can step away from your PC, make a cup of tea (or coffee!) and return to your PC and still see the above image, you have an inefficient data model.

What makes a Data Model 'good'?

Multiple parts make up a good data model. However, one area we will look at in-depth today is 'Data Hoarding'. When creating queries in Power BI Desktop, it is straightforward to pull the data sources in and not use every single table within your model. Every time the model is refreshed, regardless of whether the table is being used in your report, it will continue to load all data from that source.

In addition to this, you may use all of your tables in your report, but you may not use all of the columns within a table in your report. Again, every single row of an unused column will also be refreshed in the data model. It is not good practice to keep unused tables and columns in your model, and it is highly suggested that you de-clutter the model as often as you can to limit the amount of data being refreshed.

As part of another blog post, 'Which Power BI license is right for me?', I discuss the limitations on Pro and Premium licensing; Pro has a capped maximum limit of 1GB per data model and 10GB for Premium.

Taking the above preventative measures will allow your data models to refresh in a timelier manner and have fewer overheads for report creators and report viewers to endure.

If you would like to learn more about achieving efficient data models, please contact your Account Manager or for new enquiries, get in touch with us here.