Power BI Weekly

Issue #355 - 28th April 2026

No official announcements from the Power BI team this week, so straight into the community content.

We have a couple of data viz pieces to kick things off. Ruben Van de Voorde has written a comprehensive guide to building better bar charts in Power BI, going deep on the perceptual trade-offs behind every formatting decision. Sticking with the design theme, Juliana Smith has put together a practical font scaling guide for Power BI's new larger canvas sizes - the short version being that a bigger canvas doesn't mean bigger text (by default), and if you're not scaling font sizes proportionally to canvas height, your reports will quietly become harder to read.

Also this week, there's a fun video (in Spanish) from the community showing how to create an animated landing page in Power BI - definitely one for those who like to push the tool's design capabilities further than Microsoft probably intended.

Finally, a couple of pieces from our own blog at endjin - while not directly Power BI related, they're worth flagging for anyone involved in the data prep/engineering that sits upstream of your reporting layer. James Broome has written Medallion Architecture in Excel, which takes the familiar data engineering pattern and frames it in everyone's favourite "not a database" spreadsheet tool (stay tuned for more on this). And Barry Smart has published the results of a rigorous Fabric performance benchmarking exercise comparing Spark and Python notebooks across Pandas, PySpark, Polars, and DuckDB — the headline finding being that modern single-node engines like DuckDB and Polars can often be both faster and significantly cheaper than distributed Spark for datasets up to ~100GB. Though you also have to consider supportability, team skillsets and feature availability - all of the above being useful to understand when making infrastructure decisions in Fabric.