How Do You Translate Data Into Actionable Insights? A Power BI Weekend Experiment

-


This past weekend.


Same desk. Same drink. New mission.


After over 10 years supporting children and families through fragmented systems, I know what it looks like when missing information means missed interventions. A student's anxiety isn't flagged because systems don't talk to each other. A family's language support need falls through the cracks.


Two Worlds, One Weekend Experiment

I used Power BI to build two prototypes, combining real statistics with simulated data. The goal was testing whether the analytical approach I use in social care could reveal anything meaningful in e-commerce. I'm still learning whether that actually works.

And I'll be honest, it was genuinely fun. There's something satisfying about watching messy data transform into something clearer, even when you're not sure what you're doing yet.


The Learning Curve (Emphasis on Curve)


Here's what I'm discovering: Power BI is enjoyable, until it humbles you.

I spent Saturday morning building what I thought was a solid data model. Then I tried to create a simple measure and everything broke. One misconfigured relationship in Power Query, and the entire logic fell apart.

That moment taught me: data structure isn't just technical, it's foundational. Get the relationships wrong, and no visualization will save you. Get them right, and insight starts to flow. (At least I think that's what's happening.)

What surprised me:

The 80/20 rule seems real. 80% of my time was in Power Query building relationships. Only 20% was visualization. I'm starting to understand why that matters.

Behavioral patterns exist across contexts, but how you translate them into data-driven insight is the question. Whether analyzing student absence or customer purchasing decisions, I can see similar patterns. But structuring data to reveal those patterns meaningfully? That's the part I'm still figuring out.

Mistakes have consequences. In education, poor modeling could mean a student doesn't get support. In e-commerce, it means lost revenue. I'm learning that both matter, but one definitely carries more weight.


Prototype 1: VGS Fraværsinnsikt (Oslo High School Absence)

The Context:

In Norwegian schools, absence is often the first sign a student is struggling. But absence data alone doesn't tell you why - or what intervention might help.

What I Built

This dashboard connects absence with barriers: anxiety, learning gaps, family challenges, language support. It uses Udir statistics (2025) combined with simulated variables to explore how schools could move from reactive reporting to proactive follow-up.






The Approach I'm Testing

This tries to apply the same thinking I use in frontline work: identify the pattern (absence), understand the barrier (anxiety, language gaps), think about the intervention point (before critical thresholds). I'm curious whether this kind of behavioral analysis actually works in dashboard form.

Why It Matters to Me

This isn't about surveillance. It's about exploring whether we could catch students before they fall behind, when anxiety is still manageable, when language support can still close the gap.


Prototype 2: Domino's Pizza Operations


The Context:

I wanted to test whether my behavioral analysis approach could work in e-commerce. Could understanding customer behavior patterns across platforms reveal anything useful? Honestly, I wasn't sure.

What I Built

Using a Star Schema, I separated "Sales" from "Stores," "Platforms," and "Products." Custom DAX measures calculate Net Revenue, actual profit after platform commissions. But my real focus was on behavioral patterns: how visibility affects purchasing, how platform choice influences customer decisions.
  
    




What I noticed:

Visibility seems to drive sales , that feels like a behavioral pattern. Customer platform preferences appear to impact profitability. I'm still learning whether these patterns are meaningful or if I'm reading too much into simulated data. But it's interesting to explore.


What Might Connect These Worlds?


Behavioral Patterns Exist Across Contexts, But Translation Is the Challenge

I can see similar patterns: what barriers prevent a student from attending school, what factors influence a customer's platform choice. The behavioral lens feels similar, observe patterns, identify influencing factors, think about impact points.

But here's what I'm learning: seeing the pattern and structuring data to reveal actionable insight are two different skills. How you build relationships in Power Query, what measures you create in DAX, which visualizations actually communicate meaning, that's where the challenge lives.

Structure Seems to Matter More Than I Thought

80% of my time was in Power Query building relationships. I'm learning that bad data models lead to bad decisions, whether you're looking at schools or businesses.

Different Stakes, Similar Skills (Maybe)

The technical methods seem to transfer. Power Query, DAX, data modeling, these appear to work the same way whether you're analyzing student support needs or e-commerce profitability. But the consequences are very different: educational data directly impacts individual wellbeing. That changes how I think about every decision.




What I'm Still Learning:

Behavioral patterns exist across sectors, but translating them into data-driven insight is where the real work lives. I can observe similar patterns in student behavior and customer behavior. But structuring data to reveal those patterns meaningfully requires technical choices I'm still learning to make.

Data structure seems to determine everything. Incorrect relationships in Power Query make all downstream analysis unreliable. I'm learning this the hard way.

Most of the work happens before visualization. Power Query consumed 80% of my time. I'm starting to understand that data cleaning and relationship modeling have to come first.

Governance feels more important with real impact. When data affects vulnerable populations, transparency and data minimization feel like ethical requirements, not just best practices.

Context changes everything. The same technical skills might reveal patterns in both education and commerce. But working with student data carries weight that shapes how I think about every analytical choice.




I'm Curious:

For Domino's Pizza Norway: Does behavioral analysis of customer patterns across platforms actually work in practice? I'm genuinely curious whether this approach reflects useful decision support.

For Oslo Kommune / Utdanningsetaten: Would this kind of pattern analysis be helpful in day-to-day work, or am I missing how schools actually use data?

For anyone in data or analytics: I'm exploring roles where behavioral insights and data structure serve real decisions. If you've tried translating behavioral patterns into data-driven insight across different sectors, I'd love to hear what you learned.




Resources & References

Tools Used:

Microsoft Power BI - powerbi.microsoft.com

Power Query for data transformation

DAX for custom measures

Data Sources:

Educational statistics: Udir (Utdanningsdirektoratet) -
SSB Tabell 11771:udir.no

E-commerce and some variables in the High School data was simulated for this project.














Comments

Post a Comment

Popular posts from this blog

From Social Work to Microsoft Consultant Associate: Governing the Process of a Career Pivot.

-
Image
On 17 January 2026, I passed PL‑200! This desk. Same spot. Same time. Same drink. Every single day. Pink sticky notes = my daily minimums. Laptop = 45 minutes of focused work. No exceptions. Coming from social work, my job has always been to design systems that still work when people are tired, stressed, or having a bad day, so I applied that same mindset to learning.   Instead of relying on willpower, I built a system around my motivation, which let me catch mistakes before they snowballed. The core idea: Govern the process, not just the output. Step 1: make discipline measurable (SMART) I didn’t set a vague goal like “study more.” I needed structure and hands-on experience, so I joined the Microsoft Power Up Program (Cohort 28, Fall 2025), a free program designed to help people build practical skills through guided, hands-on learning. For three months, I deep-dived into Power Apps with one clear outcome: build an end-to-end Smart Parking & Permit Management solution whil...
Read more

The Evolution of Care

-
Image
Why I Am Building for the Frontline. As a Social Educator, I work at the intersection of municipal health, specialist services, and welfare. I see firsthand where systems break down, where data stops flowing, and where the human impact is greatest. My focus is on bridging these gaps to create insights and solutions that truly work for people. Care is not a static act. It must evolve. If my ten years in the field taught me what needs to happen, my transition into the tech world over the last six months has taught me how to make it happen at scale. Care Is Evolving - And So Am I I used to think that care only happened in a living room or a clinic. I was wrong. Care is also found in the code that ensures a patient’s history is available when it’s needed most. Care is found in the low-code interface that saves a nurse thirty minutes of redundant data entry. I am not leaving care; I am evolving into a new form of it. I am moving from direct care to systemic care . Thi...
Read more

Hierarchy Security in Dataverse

-
Image
Unlocking Hierarchy Security in Dataverse . 3-minute read | Power up journey|PL-200 Study Notes I’ve recently gained my Dataverse explorer badge from Power Up Program, and I wanted to share a specific feature that stood out to me: Hierarchy Security . It’s a documented way to solve a very common problem how to give managers visibility into their team's data without over-complicating security roles. Here is the traceable theory I’ve gathered from Microsoft Learn.  The Core Theory: Beyond Basic Roles In Dataverse, we usually think of security in terms of Security Roles (which define what actions you can take, like Read or Delete).  Hierarchy security is a documented additive layer that defines which records you can see based on your place in the organizational structure.   Verification:   It does not o verride Roles: You still need a Security Role with "Read" access to the table.   The Two Documented Models According to Microsoft documentation, there are two ...
Read more