Part-to-Whole
Icicle Chart
A rectangular, space-filling hierarchy chart where nested rectangles hang vertically to show depth and proportion.
Budget Allocation Hierarchy
Three-level breakdown: division to team to line item
View data (18 rows)
| Category | Value | Parent |
|---|---|---|
| Total Budget | 1000 | |
| Operations | 450 | Total Budget |
| Staffing | 280 | Operations |
| Salaries | 200 | Staffing |
| Benefits | 80 | Staffing |
| Infrastructure | 170 | Operations |
| Cloud | 110 | Infrastructure |
| Office | 60 | Infrastructure |
| R&D | 350 | Total Budget |
| Lab Equipment | 200 | R&D |
| Microscopes | 130 | Lab Equipment |
| Reagents | 70 | Lab Equipment |
| Software | 150 | R&D |
| Licenses | 90 | Software |
| SaaS | 60 | Software |
| Marketing | 200 | Total Budget |
| Digital Ads | 130 | Marketing |
| Events | 70 | Marketing |
Use an icicle chart when…
- Showing hierarchical breakdowns where label readability matters
- When a rectangular layout is preferred over radial sunbursts
Avoid when…
- When the hierarchy is very deep (rectangles get too narrow)
- When there is no meaningful hierarchy in the data
Data it needs
| Property | Value |
|---|---|
| Min Rows | 5 |
| Min Columns | 2 |
| Column Types | stringnumber |
| Notes | Hierarchy must be encoded as a parent reference column (or path string) — a flat string+number table is not enough. |
Visual anatomy
Marks
rectangle
Channels
width (proportion)vertical position (depth)color (category)
Axes
-
Guiding principles
Consider instead
Common mistakes
Using too many hierarchy levels, making text illegible
Not ordering children by size for easier scanning
History
Icicle plots are a partition-based hierarchy visualization introduced by Kruskal & Landwehr (1983); they share the partition algorithm with treemaps but use a layered rectangular layout that reads top-down.
Accessibility notes
Provide an indented text list; ensure each rectangle has an accessible name and value.
Related reading
Got data? Let's see what works.
Drop your CSV. You'll get an Icicle Chart plus four alternatives - ranked by which one actually fits your data best.