What is a Sankey diagram?
A Sankey diagram shows quantities in motion. Each flow is drawn as a ribbon whose width is proportional to the quantity it carries, from one stage to the next. You see at a glance what weighs, what splits and what is lost — which a table of figures never conveys as fast.
How to read a Sankey diagram
Four rules are enough to read one properly — and to spot the ones that are wrong.
Width carries quantity
That is the whole rule. A flow twice as wide represents twice as much. The eye compares thicknesses, not numbers: you see at once which of two items weighs more, without reading a single label.
Nothing is lost, nothing is created
At every node, what comes in equals what goes out. A diagram where a node emits more than it receives is either wrong or missing an input. This conservation constraint is what separates a Sankey diagram from a decorative arrow chart.
The reading direction tells a transformation
Left to right, usually: resources enter, are transformed, split, and end up as uses, losses or stocks. Losses are often the real subject — they stand out because they branch off to the side.
One unit at a time
Tonnes, terajoules, euros, cubic metres — but only one per diagram. Mixing units within a single network makes the widths incomparable and the total meaningless. It is the most common mistake.
Where the form comes from
The name comes from Matthew Henry Phineas Riall Sankey, an Irish engineer who published a diagram of a steam engine’s efficiency in 1898: it showed, for the first time with that clarity, which share of the incoming energy ended up as useful work and which share left as waste heat.
The form, however, predates him. In 1869 Charles Joseph Minard published his figurative map of the successive losses of the French army during the Russian campaign: a ribbon thinning as the army advanced and then retreated. It remains one of the most discussed graphics in the history of data visualisation — and it is a Sankey diagram before the name existed.
Since then the form has left thermal engineering for anything that can be counted and moved: materials, waste, water, money, people.
What it is used for
Energy
The original case, and still the most common: from primary energy to final uses, showing what conversions lose along the way. National energy balances are published in this form.
Materials and waste
Following a resource from one end of a supply chain to the other: what is produced, imported, processed, consumed, discarded, recycled. This is the heart of material flow analysis.
Budgets and financial flows
Where the money comes from, where it goes. A budget drawn as a Sankey diagram is understood in seconds, where a table demands a line-by-line read.
Journeys and conversions
The path of a population through stages: visitors on a site, candidates in a hiring process, patients in a protocol. Drop-offs read as losses.
How to create a Sankey diagram
You need a list of flows. For each one: where it starts, where it ends, how much it carries. One row per flow, three columns — the shape a spreadsheet naturally takes:
| Source | Target | Value (kt) |
|---|---|---|
| Harvest | Sawmill | 120 |
| Harvest | Export | 30 |
| Sawmill | Sawn timber | 72 |
| Sawmill | By-products | 48 |
From there the tool derives the nodes, places the ribbons and scales the widths. What remains is formatting — grouping, colouring, annotating — and above all the consistency of the figures, which becomes the real subject as soon as the data comes from several sources.
And if values are missing? They can often be deduced. Since a node’s inputs equal its outputs, an unknown surrounded by known values is fully determined. When several are missing at once, a reconciliation solver looks for the consistent set of values closest to the observed data, weighted by how much you trust each one. That is what separates a drawing tool from a flow analysis tool.
Free, in the browser, no account.
Common pitfalls
Too many nodes
Past forty or so nodes, the diagram turns into a plate of spaghetti and stops informing. The remedy is not to cut the data but to group it: a good tool lets you collapse a branch and expand it only on demand.
Figures that do not balance
Numbers come from heterogeneous sources, and they rarely agree on the first pass. Forcing the balance by hand, parking the gap in a “miscellaneous” item, hides the uncertainty instead of dealing with it.
Confusing it with a flowchart
A Sankey diagram is not a flowchart: it does not describe a sequence of steps or decisions, it quantifies magnitudes that circulate. If your arrows carry no quantity, it is not a Sankey.
Frequently asked questions
What is the difference between a Sankey diagram and a flowchart?
A flowchart describes a sequence of steps and decisions: its arrows indicate order. A Sankey diagram quantifies: the width of each arrow is proportional to the quantity flowing through it, and the sum of a node’s inputs equals the sum of its outputs.
Where does the name “Sankey” come from?
From Matthew Henry Phineas Riall Sankey, an Irish engineer who published a diagram of a steam engine’s energy efficiency in 1898. The form predates him: Charles Joseph Minard had used it as early as 1869 to depict the successive losses of Napoleon’s army during the Russian campaign. It is Sankey’s diagram that gave the family its name.
Can you make a Sankey diagram in Excel?
Excel has no native Sankey chart type. You can prepare the data there — one row per flow, with its source, target and value — then import it into a dedicated tool. That is the most common way of working, and the one OpenSankey accepts directly.
What data do you need to build a Sankey diagram?
At minimum a list of flows: for each one, a starting point, an end point and a quantity, expressed in a single unit. Node names are derived from that list. Everything else — colours, position, grouping — is formatting.
What if some values are unknown?
They can often be deduced. Since a node’s inputs equal its outputs, a missing value is sometimes fully determined by its neighbours. When several are missing, a reconciliation solver looks for the consistent set of values closest to the observed data, weighted by their uncertainty.
Is there a free tool to create a Sankey diagram?
Yes. OpenSankey runs free in the browser, with no account, to create a diagram, edit it with the mouse and export it as PNG or SVG.
Going further
- Starting from an Excel file — the shape to give your data, and the traps of a first import.
- Diagrams to play with — the same principles, interactive.
- The study gallery — real published flow maps.
- Data reconciliation — how we handle figures that do not balance.
- Why material flows matter — what this reading changes for a territory.