Sankey Diagram vs Alluvial Diagram: What Is the Difference?
Alluvial diagrams show how group membership changes between ordered categories or time points; Sankey diagrams show quantity flowing between nodes. See the difference with an example.
An alluvial diagram follows the same population through several ordered stages as its groups change. A Sankey diagram follows a quantity as it is split and moved between nodes. The two overlap and the names are often used loosely.
The example is closer to an alluvial diagram: 500 students are grouped by programme in year one and by outcome in year three, and the bands between show who stayed, who switched and who left. Every student appears once in each column.
A Sankey diagram does not need that structure. Its nodes can be different things in each column, and a node can lose or gain quantity, as with energy lost as heat.
Where they differ
The distinction is about what the columns are.
- In an alluvial diagram each column is the same set of entities classified in a different way, usually at a different time or by a different variable. The total is the same in every column.
- In a Sankey diagram the columns are stages of a process. Nodes can be different kinds of things, and the quantity may change along the way through gains and losses.
Why they are confused
Both draw ribbons between stacked nodes, and many tools draw both with the same engine. Some people call every such chart a Sankey, and some call every one an alluvial diagram. Choose by what you are showing: change in membership over ordered stages points to alluvial, flow and loss of a quantity points to Sankey.
Key points
- Of 500 students, 40 leave in the example: 30 from science and 10 from arts.
- Switching is nearly balanced: 20 move from science to arts and 20 from arts to science.
- Every student is in exactly one node in each column, which is what makes this alluvial rather than a general Sankey.
- The figures are invented.
The example’s data
Invented student numbers: 300 in science and 200 in arts in year 1, 270 and 190 in year 3 plus 40 who left.
| From | To | Amount (students) |
|---|---|---|
| Science (year 1) | Science (year 3) | 250 |
| Science (year 1) | Arts (year 3) | 20 |
| Science (year 1) | Left | 30 |
| Arts (year 1) | Arts (year 3) | 170 |
| Arts (year 1) | Science (year 3) | 20 |
| Arts (year 1) | Left | 10 |
The same data as CSV
source,target,value
Science (year 1),Science (year 3),250
Science (year 1),Arts (year 3),20
Science (year 1),Left,30
Arts (year 1),Arts (year 3),170
Arts (year 1),Science (year 3),20
Arts (year 1),Left,10How it works
Ask what the columns are
The same people sorted differently at each stage: alluvial. Stages of a process with gains and losses: Sankey.
Check conservation
If every column has the same total, an alluvial diagram fits. If the total changes, use a Sankey.
Use the same data format
Both are a source, target, value list, so you can start with either and switch.
Frequently asked questions
Is an alluvial diagram a type of Sankey diagram?
Many authors treat it as a special case: a Sankey-style diagram where each column partitions the same population. Others treat the two as separate charts. The practical difference is as described above.
Which should I choose for a customer journey?
If you follow the same customers through stages and care about who changes group, alluvial. If you also want to show where people drop out, a Sankey with a node for each drop-off.
Edit this diagram
Open the example as your own chart and replace it with your categories.