Flow, process and networks
Sankey diagram
Shows quantities transferred between nodes using link widths proportional to flow.
- Diagram
- Flow
- Composition
What it answers
Where does a quantity flow and how much takes each route?
- Shape
- Nodes joined by width-scaled links
- What it shows
- Major flows, splits and joins.
Example
The question
Where do website visitors come from, and how many end up buying?
Website visits by source, landing page and outcome, September (thousands)
Illustrative data. It belongs to no company.
When to use it
Use it when
A manageable set of additive transfers with consistent units.
Avoid it when
Too many crossing links or unexplained imbalances.
How to read it
Follow links from sources to destinations and compare widths using the value labels.
Note
Define units and explain sources, sinks or non-conservation.
The data you need
- Layout
- Edge list plus optional node table
- Fields
- Source: text ID; Target: text ID; Value: nonnegative number
- One row is
- One directed source-to-target quantity. What is the grain?
- Rules
- Use consistent node IDs and units. Aggregate duplicate pairs deliberately. Check balance where flow conservation is expected.
- Optional
- Separate node table: NodeID,Label,Group.
- Preparation
- Resolve node identities and endpoint joins; define direction and duplicate/self-link handling. Aggregate weights only under a stated relationship or flow definition.
- Units
- Unweighted links encode presence. Weights need a unit; flow shares divide by stated source, destination or network totals, not an assumed common denominator.
Worked example
Flows A->B30, A->C20, B->D30. A sends 50; B passes its 30 to D. Link widths represent those quantities.
| Source | Target | Value |
|---|---|---|
| A | B | 30 |
| A | C | 20 |
| B | D | 30 |