Difference between creating a dataset for yourself or creating for others

When I started working with Power BI, all the datasets I created was meant to support the report I was building. As more of my colleagues started embracing Power BI, they wanted to create their own reports also. When I shared the datasets, most of the time they ended up not using it, although it contained all the information they needed.

I started to realize that the reason that what they saw in my dataset is not something that can help them achieve their goals faster but a big mess that they needed to understand before they start creating their reports. It was just simpler for them to start over and build a new one.

I did not saw that my datasets had any problems, because I was the one creating them. All the transformations, renaming, aggregations, filters, relationships, bridge tables, calculations… were done by me, so it was easy for me to understand it. I know all the potential issues, which measure works with which dimensions, etc. But for them it was just a mess.

With the addition of Personalized visuals, which enables end users to modify the visuals in a report without been an author, the quality and usability of the datasets became much more important than ever.

Why is this important?

Datasets in Power BI contain a most important part of any analytics solution. The data model with the calculations.

The ability to reuse datasets is one of the key ways to eliminate unnecessary duplications of effort and data. It also helps you get closer to achieve a single source of truth.

One good dataset can be a source of 10s or even 100s of reports. If you think about maintenance and change management than it is obvious why having a smaller number of good quality, optimized and clean datasets is better.

Ideas to improve the data model

  • Make column names easy to understands and use business lingo
  • Hide unnecessary columns, measures, and tables
  • Organize columns and measures with folders
  • Provide descriptions to measures (add the DAX formula at least) and columns that need explanation
  • Make sure that additions filters are not needed to display the right values using a measure. For example, the need to filter out returned sales in the visual filter to give you the proper sales volume, have a measure that already provides that
  • Create Layouts for each Fact table in the Model view


If you think your data model will be used by someone other than you, make sure it is as easy to use as you can possibly make it.

The ability to reuse datasets is one of the most under used feature of Power BI.

Please follow and like us:

About the author

Mihaly Kavasi

Data-driven Decision-Making Enthusiast.
Power BI Expert and Trainer. Helping organizations and communities to use data effectively.
Big fan of RE and Electric cars.
Working @ Avanade

View all posts