By Jeremiah Willett
Data Principles is currently building a training course for a client: Introduction to Power BI. One of the most important topics in the course is data modeling in Power BI. Many introductory training courses to Power BI start with building the visuals or Data Analysis Expression (DAX) tricks. However, at Data Principles, we believe that building visualizations in a BI tool (Qlik, Power BI, Tableau, etc.) should be the final step in the data storytelling process. First, and most importantly, we tell a story with our data model.
The problem with starting with the drag-and-drop visuals and DAX tricks when teaching Power BI to a student is that there will be no foundational understanding of the data and how each data element relates to another. This isn’t just a problem of not knowing something; building visuals without a complete data model will ultimately lead to the dashboards breaking and wasting time.
An example case that I used in the training is a political campaign analytics dataset. We had various tables with campaign-related data: staff activity, donations, events, and voter outreach activity. Each of these campaign elements interacts with the other, and that is the story
that we want to tell in our data model.
Every dataset has a narrative structure. Dimensions are the nouns of our story: Campaigns, Staff, Donors, Dates, Events. Facts are the verbs or actions: Donations, Volunteer Hours, Doors Knocked, Event Attendance, Donor Phone Calls. When the nouns and verbs in the dataset are organized properly, the business questions will flow naturally:
- “Which EVENTS drive the most DONATIONS?”
- “Which outreach channels convert best?”
- “How does volunteer time correlate with campaign performance?”
A well-built data model will tell the story automatically. When our Donation fact table has a relationship to our Party (Donors) dimension table, our Campaign dimension table, and our Date dimension table, we can immediately see who gives, when, and to which campaign. The same applies to each and every table from our campaign dataset. The relationships between the various tables in your dataset will create a narrative before a single visual is built.
Conclusion
A beautiful visual that shows Total Donations by Campaign by Month in a bar chart, which will be immediately useful to the leadership in your organization, is only possible if the relationships between each fact table and dimension table are properly defined. Always start with the data model before building your visuals: this will save an enormous amount of time and will ensure that the stories that you will tell in your dashboards are accurate and useful to your organization.
Jeremiah Willett, CDMP: Empowering Smarter Decisions with Data
He is an Associate Manager of Data Engineering at Data Principles, where he helps organizations turn complex data into clear, actionable insights through innovative business intelligence solutions. With experience across SQL, Qlik Sense, data integration, and project management, he is passionate about empowering organizations to make data-driven decisions. Jeremiah also serves as the Vice President of Finance for DAMA Phoenix and holds a Certified Data Management Professional (CDMP) credential.
Join Our Data Community
At Data Principles, we believe in making data powerful and accessible. Get monthly insights, practical advice, and company updates delivered straight to your inbox. Subscribe and be part of the journey!
