Sankey Diagram

What is a Sankey Diagram?

A Sankey diagram is a flow visualization where the width of arrows (links) is proportional to the quantity being transferred. Nodes represent stages, categories, or entities, while the flowing bands show how quantities move, split, merge, or transform between them. This makes Sankey diagrams ideal for understanding resource flows, process stages, and distribution patterns.

Why Use Sankey Diagrams?

Sankey diagrams excel when you need to:

• Track resource flows: Visualize how materials, money, energy, or information moves through a system.
• Identify major pathways: Instantly see which flows dominate by their visual weight.
• Spot leakage or loss: Detect where quantities diminish as they flow through stages.
• Understand distribution: See how a source splits into multiple destinations.
• Compare proportions: The width encoding makes relative quantities immediately apparent.

How It Works in GraphPolaris

Creating and exploring a Sankey diagram in GraphPolaris is straightforward:

1. Load your flow data containing sources, targets, and quantities.
2. Generate the diagram with automatic node positioning.
3. Explore flows: Hover over links to see exact values, click nodes to highlight connected flows.
4. Filter and focus: Isolate specific pathways or filter by flow magnitude.

GraphPolaris provides multiple layout options and supports multi-level flows with intermediate nodes.

Visual Patterns

Understanding common visual patterns in Sankey diagrams helps you quickly interpret flow dynamics. Here are the key patterns to look for:

Dominant Flow

A single thick band that carries most of the quantity indicates a dominant flow—the primary pathway through the system.

    ╔══════════╗                    ╔══════════╗
    ║  Source  ║════════════════════║   Main   ║
    ╚══════════╝        ↑           ╚══════════╝
         │          Dominant
         └──────────────────────────┐
                                    ↓
                               ╔═════════╗
                               ║  Minor  ║
                               ╚═════════╝

What to look for: One band significantly wider than others. The dominant flow often represents the "happy path" or primary use case.

Splitting (Distribution)

When a single source divides into multiple destinations, the bands fan out showing how the total is distributed.

                               ╔═══════════╗
                          ┌════║  Dest A   ║
    ╔══════════╗         ═╪════╚═══════════╝
    ║  Source  ║══════════╪    ╔═══════════╗
    ╚══════════╝         ═╪════║  Dest B   ║
                          ╪════╚═══════════╝
                          │    ╔═══════════╗
                          └════║  Dest C   ║
                               ╚═══════════╝

What to look for: One node with multiple outgoing bands. Compare widths to understand distribution proportions.

Merging (Aggregation)

When multiple sources combine into a single destination, the bands converge showing how quantities aggregate.

    ╔═══════════╗
    ║  Src A    ║════╗
    ╚═══════════╝    ╪═══╗
    ╔═══════════╗    ╪   ║    ╔══════════╗
    ║  Src B    ║════╪═══╬════║  Total   ║
    ╚═══════════╝    ╪   ║    ╚══════════╝
    ╔═══════════╗    ╪═══╝
    ║  Src C    ║════╝
    ╚═══════════╝

What to look for: Multiple incoming bands joining at a single node. The output width equals the sum of inputs.

Funnel (Progressive Loss)

When flow diminishes at each stage, the diagram shows a funnel pattern—common in conversion processes where quantity is lost at each step.

    ╔═══════════════╗
    ║    Stage 1    ║
    ╚═══════════════╝
           ║║║
           ║║╚══════► Loss
           ║║
    ╔════════════╗
    ║   Stage 2  ║
    ╚════════════╝
          ║║
          ║╚═══════► Loss
          ║
    ╔═════════╗
    ║ Stage 3 ║
    ╚═════════╝

What to look for: Progressively narrowing bands from left to right. The difference at each stage represents loss or dropout.

Balanced Flow (Conservation)

When total input equals total output at each stage (no loss), flows maintain consistent total width—indicating a closed or conserved system.

    ╔══════════╗         ╔══════════╗         ╔══════════╗
    ║  Input   ║═════════║ Process  ║═════════║  Output  ║
    ╚══════════╝    ↑    ╚══════════╝    ↑    ╚══════════╝
                    │                    │
              Same width            Same width

What to look for: Total band width remains constant across stages. What goes in must come out.

Feedback Loop

When some flow returns to an earlier stage, it creates a visible loop pattern—common in recycling, rework, or iterative processes.

    ╔══════════╗         ╔══════════╗         ╔══════════╗
    ║  Start   ║═════════║ Process  ║═════════║   End    ║
    ╚══════════╝         ╚══════════╝         ╚══════════╝
         ▲                    ║
         ║                    ║
         ╚════════════════════╝
              Feedback/Rework

What to look for: Bands that flow backward (right to left) or loop back to earlier nodes.

Bottleneck

When many wide inputs converge into a narrow output stage, it indicates a bottleneck—a constraint limiting throughput.

    ╔═══════════╗══════╗
    ║  Input A  ║      ║
    ╚═══════════╝      ║     ╔═════════╗
    ╔═══════════╗      ╠═════║Bottlenck║═══►
    ║  Input B  ║══════╣     ╚═════════╝
    ╚═══════════╝      ║          ↑
    ╔═══════════╗      ║      Narrow
    ║  Input C  ║══════╝
    ╚═══════════╝

What to look for: Many thick incoming bands converging to a thin outgoing band. Indicates capacity constraints.

Multi-Level Cascade

Multiple stages where flows split and recombine, showing complex pathways through a system.

    ╔═══════╗     ╔═══════╗     ╔═══════╗     ╔═══════╗
    ║  Src  ║══╦══║ Mid A ║══╦══║ Mid X ║══╦══║ Dest  ║
    ╚═══════╝  ║  ╚═══════╝  ║  ╚═══════╝  ║  ╚═══════╝
               ║  ╔═══════╗  ║  ╔═══════╗  ║
               ╚══║ Mid B ║══╩══║ Mid Y ║══╝
                  ╚═══════╝     ╚═══════╝

What to look for: Multiple columns of nodes with flows crossing between them. Trace individual pathways through the stages.

Sankey diagrams provide intuitive flow visualization—and with GraphPolaris, you can interactively explore complex flow networks to understand how quantities move through your systems.