My notes
These are notes from training sessions, events and conversations with teams. They are not summaries or polished articles. They are short reflections on topics that stayed with me long enough to be worth writing down.
Most recent notes
-
Four questions before you build: the source under your report
Before you build a report you need four answers about its source in Azure: where it sits, what it consists of, who has access and who gets the cost mail. -
A budget in Azure: what it does and who gets the mail
A budget in Azure sends a mail at a threshold and stops nothing. Your team's scheduling tool shows when that mail arrives, which costs count and who should get it. -
Storage, compute and networking: what can you take separately?
Storage, compute and networking sound like one family, but two of them you can order separately and one you never can. Your team's scheduling tool shows the difference.
Find your topic
Every note belongs to one topic. Jump straight to it, or type a word to search all notes.
No notes found. Try a different word.
Practical Power BI tips 13 notes
Directly applicable to your own report, from DAX basics to the filter-context series. Read them in order: each note builds on the one before.
-
How much, how often, or how many distinct: choosing the right DAX aggregate
Learn when to reach for SUM, COUNTROWS, or DISTINCTCOUNT by getting the question right first. Day 3 of the DAX basics series for Power BI. -
DIVIDE in DAX: the cleaner division for your Power BI measures
Every division in a Power BI measure belongs in DIVIDE, not slash. What the function does, why it keeps your reports clean, and how to remember the argument order. -
DAX filter context: the measure has no row
A measure in DAX has no idea where it lives. It sees the table that filter context leaves behind. A practical guide with one anchor measure across three contexts: visual, slicer, and row. -
CALCULATE replaces the filter on that column: day 1 of the DAX filter series
CALCULATE is the only DAX function that puts filtering inside the measure itself, and that explains why your year slicer can seem to stop working. -
FILTER in DAX: when to use it and when to let CALCULATE do the work
FILTER is an escape hatch in DAX, not the default. When to use a Boolean argument in CALCULATE, and when FILTER is unavoidable. -
REMOVEFILTERS and ALL in DAX: knowing when to use which
Day 3 in the DAX filter burst: REMOVEFILTERS and ALL. Inside CALCULATE they do the same thing, but ALL has a second role as a table function. -
ALLEXCEPT in DAX: hold one grouping, let the rest go
Day 4 of the DAX filter burst: ALLEXCEPT wipes every filter on your table except the columns you name, so your share within category keeps adding up. -
ALLSELECTED in DAX: your denominator follows the user's selection
Day 5 of the DAX filter burst: ALLSELECTED clears the filter your visual sets and keeps the rest of the selection intact, so your percentage matches the screen again. -
The DAX filter cheat sheet: five functions, two jobs
Five DAX filter functions on one scale: what CALCULATE, FILTER, ALL, REMOVEFILTERS, ALLEXCEPT and ALLSELECTED do, and how much stays in your denominator. -
Power BI Desktop: saving to OneDrive and SharePoint stops in October
From October, a Power BI Desktop version from March 2026 or older can no longer save to or share through OneDrive and SharePoint. How to check your version in thirty seconds. -
TMDL view on the web: your model as readable text in the browser
Since the July update you can read your published Power BI model as plain text in the browser. What the TMDL view returns, and when Desktop is still the tool. -
Refreshing in Power BI: schema, data, or both
Since the August update, refreshing means choosing between schema, data or both. What your model stores, and when to reach for which option. -
The Power BI date picker: calendar month or rolling window
Since the August update the date picker locks a rule into your slicer. The period type decides whether you get a calendar month or a window that ends today.
Data modelling 5 notes
Why the model comes before the measure: what DAX actually does, measures versus columns, and the foundation under every calculation.
-
Copilot agents on your BI: four spots in your model and the RLS check before you start
Data Agents in Microsoft 365 Copilot are generally available. Four spots in your semantic model to prepare, plus the RLS check you cannot skip. -
What DAX actually does: three levels in every Power BI report
Get data in Power Query, model in the data model, calculate in DAX. The three-level split is the foundation of every Power BI report. -
Mise en place for DAX: measure or calculated column?
Measure or calculated column? Björn explains the timing question behind DAX's most common beginner stumble. Part 2 of the DAX basics series. -
Before you write a measure: is your model right?
When a measure behaves unexpectedly, the problem is almost never your DAX. Three places in your Power BI model to check before you touch CALCULATE. -
The description field on your measure: four sentences that make your number explainable
Power BI fills in a measure's name, formula and formatting for you, but not its meaning. The description field is the only place that agreement can live.
Agentic Power BI 2 notes
AI agents alongside your BI work: from your first agentic report to building an agent in Foundry.
-
Build your first agentic Power BI report in five steps
Get started with agentic development for Power BI on your own laptop. What you need, why skills matter, and a five-step plan to your first run. -
Building an agent in Foundry: five times adding something human
Five settings in Microsoft Foundry give an agent character, hands, memory, conscience and a mirror. The practical guide for AI-901 domain 2.
Certification & growth 8 notes
The Microsoft exams and what they mean for your learning path: AZ-900, DP-900, AI-901 and beyond.
-
AI-901 is the new Azure AI Fundamentals baseline: what changes and what to do about it
AI-900 retired on 30 June 2026. Here's what AI-901 actually tests, what stays the same, and which step makes sense depending on your situation. -
The responsible AI block in AI-901: from six principles to four questions
The responsible AI block in AI-900 and AI-901 has six official principles. Björn Kats explains how to turn them into four practical questions you ask before every AI use. -
Three corners of one landscape: which Microsoft fundamentals do you take first?
AZ-900, DP-900, and AI-901 are not a ladder. They cover three corners of the same landscape. Start at the corner closest to your daily work. -
Cloud, tenant, resource and subscription: the four questions behind one word
Cloud, tenant, resource and subscription sound like synonyms, but they answer four different questions. One campaign website shows exactly why that difference matters. -
Authentication and authorization: two doors between you and that resource
Authentication and authorization look like one access request, but they are two different doors. The screen your colleague sees tells you which one is shut. -
Storage, compute and networking: what can you take separately?
Storage, compute and networking sound like one family, but two of them you can order separately and one you never can. Your team's scheduling tool shows the difference. -
A budget in Azure: what it does and who gets the mail
A budget in Azure sends a mail at a threshold and stops nothing. Your team's scheduling tool shows when that mail arrives, which costs count and who should get it. -
Four questions before you build: the source under your report
Before you build a report you need four answers about its source in Azure: where it sits, what it consists of, who has access and who gets the cost mail.
Notes from the Power BI User Days 2026
At the Power BI Gebruikersdagen 2026 I attended several sessions on modelling, governance, AI and the evolution of the platform. Below are the notes from that event.
-
Data governance
Why governance is becoming more relevant, what ownership means in practice, and how to get started as an organisation. -
AI and Power BI
Three directions in which AI is changing the platform: MCP servers, data agents in Fabric and building custom visuals with code. -
Data products and data contracts
What it means to treat data as a product, and how data contracts give data users predictability. -
AI now builds your data model too
How you can use instructions, skills and an MCP server to have AI generate semantic models.
Other notes
-
From tables to models
Pivotal data moment #3: build a small star schema yourself. Pick a domain, click the columns into place and see the difference between facts, filters and keys. -
Frequently Asked Queries: Power Query
Pivotal data moment #2: a nod to Power Query. How to start, which functions are inside, where you'll find it and what to do when you want to go further. -
From VLOOKUP to join
How a simple Excel trick showed me for the first time how tables connect and everything that joins brought after that. -
PL-300: Why fundamentals still matter in the Fabric era
FabCon session / April 2026. Fabric brings more workloads. The end goal has not changed. A look at why the four pillars behind PL-300 still offer the clearest path. -
Starting small
Why you do not need a large data team or complex infrastructure to build useful reports. -
Data modelling
Why a solid data model is still the most important Power BI skill, and why shortcuts make things harder later. -
Power BI as a development platform
How Power BI has grown into a development platform with Git, CI/CD pipelines and code integrations.
About these notes
The topics covered here, governance, AI, modelling, the role of fundamentals in a broadening toolset, are the same themes that come up regularly when working with teams on training and custom sessions. Writing them down is a way of thinking through them more carefully.