Flow, process and networks

Alluvial diagram

Shows how group quantities are distributed between categories at successive stages.

  • Diagram
  • Flow
  • Composition

What it answers

How do group memberships change across stages?

Shape
Category blocks linked across stages
What it shows
Group continuity and redistribution.

Example

The question

How did customers move between plans over three years?

Customers by plan, 2023 to 2025 (alluvial diagram, hundreds)

Customers by plan, 2023 to 2025 (alluvial diagram, hundreds)Three stacked columns for 2023, 2024 and 2025 with bands showing customers moving between plans. Basic shrinks from 55 to 34 hundred; Plus grows from 34 to 40; Pro grows from 11 to 26.202320242025Basic 55Plus 34Pro 11Basic 44Plus 37Pro 19Basic 34Plus 40Pro 26
Customers by plan, 2023 to 2025 (alluvial diagram, hundreds)Three stacked columns for 2023, 2024 and 2025 with bands showing customers moving between plans. Basic shrinks from 55 to 34 hundred; Plus grows from 34 to 40; Pro grows from 11 to 26.202320242025Basic 55Plus 34Pro 11Basic 44Plus 37Pro 19Basic 34Plus 40Pro 26
Three stacked columns for 2023, 2024 and 2025 with bands showing customers moving between plans. Basic shrinks from 55 to 34 hundred; Plus grows from 34 to 40; Pro grows from 11 to 26.

Illustrative data. It belongs to no company.

When to use it

Use it when

Categorical transitions where complete paths matter.

Avoid it when

When only unrelated stage totals are known.

How to read it

Follow bands through ordered category axes and compare band widths.

Note

Conventions overlap with Sankey diagrams.

The data you need

Layout
Wide path-count table
Fields
Stage1,Stage2...: category text; Count: nonnegative number
One row is
One category path across stages with its frequency. What is the grain?
Rules
Declare stage order and aggregate identical paths. Keep counts nonnegative.
Optional
Raw alternative: EntityID,Stage,Category. Edge-only data may lose complete path identity.
Preparation
Validate the input grain and missing values. Declare stage order and aggregate identical paths. Keep counts nonnegative.
Units
Relationships and positions have no numeric unit by default. Label any counts, durations, probabilities or weights separately.
Example of the input layout: Wide path-count tableA wide table: one row per item, with a separate column for each measure or period.Stage1Stage2...Count…ABC
A wide table: one row per item, with a separate column for each measure or period.

Worked example

Paths A-B-D30 and A-C-D20 start with 50 in A. At stage 2,60% pass throughB and 40% throughC.

Stage1Stage2Stage3Count
ABD30
ACD20