Power BI Customization

Customizing a Sankey Diagram in Power BI

Customise the colours, labels, formatting, and behaviour of a Sankey Diagram in Power BI — formatting pane options, theme overrides, conditional formatting, source-code edits.

Once you have a Sankey Diagram on your Power BI report, the next question is always: how far can I customise it? The answer depends on whether you're working with a built-in chart (limited; only what the formatting pane exposes), a marketplace custom visual (limited to whatever options the vendor exposed; paid tiers often unlock more), or an AI-generated visual (every property is exposed because you own the source).

This guide walks through the four customisation surfaces in order from most accessible to most powerful: the formatting pane, conditional formatting, theme overrides via JSON, and source-code edits. The first three apply to every Power BI custom visual; the fourth applies only when you have the TypeScript source — which you do whenever you generate a visual with VizForge.

Step-by-step

1
Start with the formatting pane
Click the visual, then the paint-roller icon. Every option the visual exposes appears here — colours, font sizes, label positions, axis behaviour, cross-filter behaviour. For an AI-generated sankey diagram, the formatting pane is whatever you described as a settings option in the original prompt.
2
Apply conditional formatting where supported
Right-click most colour or label-text properties → Conditional formatting. Drives the property from a DAX measure or rule. Useful for variance-driven colour rules (red/green based on threshold), KPI-status text, or data-driven labels.
3
Override via report theme JSON
Power BI report themes (.json) can set defaults for any custom visual property. View → Themes → Browse for themes. Themes are how enterprise design systems propagate across many reports without per-visual edits.
4
Edit the TypeScript source for deeper changes
When the formatting pane doesn't expose what you need (e.g. add a new bar-rounding option, change the cross-filter behaviour, integrate with a different DataView shape), open the visual.ts file in VS Code. The source ships with every VizForge-generated .pbiviz; rebuild with `pbiviz package` after edits.
5
Re-prompt for substantive redesigns
If the change is large enough that source edits feel laborious — different chart family, completely different layout — describe the new vision and re-generate rather than refactor. Generation is fast enough that re-prompting beats refactoring for changes above a certain size.

The AI alternative: Need a sankey diagram customised exactly to your spec? Describe what you want — VizForge generates it with every option you mentioned exposed in the formatting pane. Source code yours.

Generate a sankey diagram on VizForge →

When to use a sankey diagram

Customer journey: acquisition channel → onboarding step → conversion or churn
Budget cascade: total revenue → segment → cost bucket → net margin
Energy / material flow audits (sources → intermediates → outputs)
Multi-stage qualification funnels where candidates flow between named states
Employee headcount movement (joins, exits, transfers between departments)

Frequently asked questions

What if the formatting pane doesn't have the option I need?

Two paths. (a) Re-generate with a refinement prompt asking for that specific option to be exposed. (b) If you have the TypeScript source, add the option directly to settings.ts and the formatting pane reflects it on next package.

Can I keep the same .pbiviz across multiple reports?

Yes — once imported, the .pbiviz is a Power BI org-visual you can re-use. Or push it to your tenant's Org Visuals so anyone in your org can use it without importing.

Will customisations survive a regeneration?

Source-code edits won't carry across a from-scratch regeneration. Use Refine instead — it produces a minimal diff on the previous version, preserving everything not directly mentioned in your refinement prompt.

Does the Sankey work in Power BI Service or only Desktop?

Both. Generate the .pbiviz, import it into Power BI Desktop, build your report, publish to Service — the visual renders in the browser exactly as it does locally.

Can a Sankey cross-filter other visuals on the page?

Yes. Every VizForge-generated Sankey includes SelectionManager integration, so clicking a node or link acts as a slicer on the rest of the page.

How many levels can the Sankey handle?

Arbitrary depth. The AI-generated visual infers levels from your source/target columns — 2 levels, 5 levels, 12 levels — and lays them out automatically.