How to Read a Sankey Diagram: A Short Guide With an Example
Learn to read a Sankey diagram: what band width means, how to follow a quantity left to right, how to spot losses and what these diagrams cannot tell you.
Read a Sankey diagram by following the widest bands. Their width is the quantity, so where a band is thick, a lot is moving.
The example follows a reservoir's water to its uses. Most of it goes to agriculture, some to households and a little to industry, and part of the households' share is lost to leaks.
Reading it takes three moves: find where the quantity starts, follow the thickest bands, and look at what is lost or left over.
Three moves
- Start at the left. That is where the quantity enters. Here it is 1,000 megalitres from one reservoir.
- Follow the widest band. Agriculture is the thickest, so it is the biggest user.
- Look for what is lost or left over. The Leakage node is water that is not used by anyone, and it is the one worth asking about.
What a Sankey diagram does not tell you
It shows proportion, not cause. It does not say why leakage is 100 or whether that is high. It also hides time: one diagram is one period, and two diagrams for two periods need to use the same scale to be compared.
Thin bands are easy to miss. If a small flow matters, label it or move it into its own node.
Colour and order
Colour usually marks the source of a band or the group of a node, not an amount. The vertical order of nodes within a column is a layout choice and carries no meaning unless the author says it does.
Key points
- 1,000 megalitres leave the reservoir: agriculture 450, households 400, industry 150.
- A quarter of the households' water (100 of 400) is lost to leakage.
- Agriculture and industry end at their own node, which means this diagram does not follow what happens to that water afterwards.
- The figures are invented.
The example’s data
Invented volumes in megalitres. Not data about any real reservoir.
| From | To | Amount (megalitres) |
|---|---|---|
| Reservoir | Households | 400 |
| Reservoir | Agriculture | 450 |
| Reservoir | Industry | 150 |
| Households | Used | 300 |
| Households | Leakage | 100 |
The same data as CSV
source,target,value
Reservoir,Households,400
Reservoir,Agriculture,450
Reservoir,Industry,150
Households,Used,300
Households,Leakage,100How it works
Find the starting node
It has bands leaving it and none entering.
Follow the widest bands
They carry most of the quantity.
Check the losses
Look for nodes with no way on, and ask whether they are an outcome or a leak.
Frequently asked questions
Does the vertical position of a node mean anything?
Usually not. Position within a column is chosen to reduce crossings. Only the column, the node height and the band width carry meaning.
Can I compare two Sankey diagrams?
Only if they use the same unit and the same scale. Otherwise the same width means different amounts.
Open this diagram and explore it
Open the example as your own chart and change an amount to see what moves.