Forum Discussion
Hide header row in table visual and set column widths
- 1 year ago
Hi Anon9876 , Thank you for reaching out to the Microsoft Community Forum.
No, Power BI does not currently allow you to force column headers to wrap independently of column width. Header text will only wrap when it exceeds the width of the column and there is no option to manually insert line breaks or enable force wrap behaviour. If the column is wide enough to display the entire header in one line, Power BI will not wrap it, even if wrapping had occurred previously when the column was narrower. The only way to restore wrapping is to reduce the column width until the text naturally wraps again.
If you're committed to using table visuals and want a vertically stacked layout that looks seamless, your current method of disabling auto-size, applying temporary wide number formatting to align columns and using formatting tricks to hide headers remains the only viable option in Power BI today. This approach, as described by alish_b , is valid for static reports where layout precision is essential and interactivity or filtering isn't central.
However, if you're open to restructuring your model for a more maintainable and dynamic report, danextian 's suggestion is the better long-term approach, using a matrix visual with either a disconnected table or calculation groups. This lets you display measures like Current Month and Year-To-Date as row items, eliminates the need for layered visuals or header hacks and aligns better with Power BI's native capabilities for grouping and formatting. It does take some setup, but it’s far more scalable and robust for evolving reporting needs.
hi Anon9876
I wouldnt' follow your current route. Column headers cannot be entirely hidden - you can change the font size, color or replace the name in the visual with a space but they will still remain in the headers. It is also difficult to perfect align the column headers - while you can disable auto-resize, the problem is if the current width is not enough to accomodate and increate in the lenght of the figures, you will end up with either truncated or wrapped values. Instead, I would make use of calculation groups to apply the same time intelligence logic to different measures and a matrix viz that can be modified to look like a table