- A Sankey diagram shows how a measured quantity moves between connected stages, categories, or destinations.
- Flow width usually represents value, with wider streams indicating larger quantities.
- A Sankey chart can be useful when one shared measure moves in a clear direction and you need to show where it splits, combines, or ends.
- Check units, labels, and methodology before comparing flows. A visually wider stream may not always represent an exact difference.
The U.S. Department of Energy defines a Sankey diagram as a directional flow chart whose stream width is proportional to flow quantity. Connected bands show how that quantity moves from one point to another, including major paths, splits, and endpoints.
That visual pattern is useful when your question involves movement. Where did customers go after starting a trial, and how was a budget distributed? Which supply chain stages received the most material?
This guide explains how to read the chart, prepare its data, and decide when another format would tell the story better. For broader context, explore this data visualization guide.
What Is a Sankey Diagram?
A Sankey diagram is a directional visualization of a measured quantity flowing from sources to destinations. The path may pass through intermediate stages, letting you follow a process across several steps without losing sight of the total volume.
The rectangles or labels are called nodes. The bands connecting them are links or flows. When the chart uses a quantitative scale, link width encodes the value moving between two nodes.
For example, a customer journey might begin with trial sign-ups, split into activated and churned users, then continue from activated users to paid subscriptions. The diagram makes the size of each path visible while preserving its direction.
The U.S. Department of Energy explains that line widths indicate volume in its manufacturing energy tool. That same reading rule can apply to money, materials, customers, or other consistent measures.
A useful Sankey diagram should do more than produce an attractive shape. It should help you spot a flow worth investigating, then ask a sharper follow-up question about the underlying data.
How to Read Nodes, Links, and Flow Widths
Start at the source node, then follow each link in its displayed direction. A destination node receives the flow, while an intermediate node can both receive and send volume as the process continues.
Next, compare widths within the same chart. A wider link may mean a larger value when flows use the same unit, timeframe, and scale. Before comparing precisely, check the labels, legend, and methodology.
For a dense chart, consider tracing one source at a time. Note its outgoing links, follow each destination, and compare only the paths that answer your question. This approach can reduce the chance of mixing unrelated stages or units.
Color may distinguish stages, categories, outcomes, or another grouping. Do not assume it represents performance unless the legend says so. Labels should identify each node and, when space permits, show the value or percentage associated with the link.
Before treating flow widths as precise measurements, check the chart's legend, units, and methodology.
When Should You Use a Sankey Diagram?
A Sankey diagram can be useful when you need to trace one shared measure from A to B. Where did it split, combine, or finish? Look for a meaningful direction and a consistent unit across links.
Potential Sankey diagram use cases include customer journeys, conversion paths, budgets, supply chains, and energy or material flows. You may also consider one for movement between customer segments, product tiers, channels, or operational stages.
Imagine a fictional dataset with 10,000 trial sign-ups. A Sankey chart could show 7,200 moving to activation and 2,800 moving to churn. Another flow could show 1,600 activated users moving to paid status.
In this fictional example, the chart shows recorded movement between stages. It does not show why people activated, churned, or purchased.
Before choosing the format, define the question in one sentence. If the question centers on movement and proportional volume, a Sankey diagram may fit. If it centers on rank, trend, or a simple stage-by-stage rate, choose the right chart type for that task instead.
When a Different Chart Is Clearer
A Sankey diagram may be harder to read when there is no natural flow or links use different units. Dozens of small categories can create many crossings. Circular or bidirectional relationships may also make direction less clear.
Consider a bar chart for precise side-by-side category comparisons. A line chart may suit change over time, while a funnel can summarize a simple ordered conversion sequence. Consider a network view when connections matter more than measured directional volume.
Clutter is another reason to reconsider the format. If readers must trace many similar links or decode unexplained colors, consider grouping minor flows or switching formats to avoid misleading charts.
How to Prepare Data for a Sankey Diagram
Many Sankey implementations use a link table with source, target, and value fields. Each row can represent one directional connection, while the value describes how much of the shared measure travels along that link.
You may also add a stage, category, color group, or label field. Keep those additions tied to a clear reading task. Extra fields should help readers interpret the flow, rather than add decoration.
For width comparisons, consider using the same unit and timeframe across values. If one row counts customers and another records revenue, avoid placing both on one quantitative scale.
Consider whether incoming and outgoing totals should reconcile at intermediate nodes. Some processes may include losses, gains, retained volume, or data exclusions. Document any known difference instead of forcing totals to match.
A quick reconciliation note can help you review each intermediate node. Record its incoming total, outgoing total, and any defined remainder. The review may highlight missing links, duplicated records, or inconsistent filters before you create the visual.
Review definitions before you build the chart. Confirm what each stage means, how duplicates are handled, and whether a person or item can appear in more than one path. ThoughtSpot's product visualization page describes interactive visualizations with 25+ native chart options, but it does not state native Sankey diagram support.
A Simple Sankey Data Example
The following fictional trial-activation flow shows a compact source-target-value structure:
| Source | Target | Value |
|---|---|---|
| Trial | Activated | 7,200 |
| Trial | Churned | 2,800 |
| Activated | Paid | 1,600 |
The first two rows account for all 10,000 trial sign-ups. The third row shows that 1,600 activated users later moved to paid status. It does not describe the remaining activated users unless the dataset includes their paths.
This fictional table does not show why users moved between stages. It only summarizes the paths and values included in the example.
How to Make a Clearer Sankey Diagram
Consider starting with the decision or question the chart should support. Write down the measure, unit, timeframe, source stage, and destination stage before choosing colors or arranging nodes.
You can then prepare a source-target-value table and select stages that serve the question. Consider direct labels, readable values, and a direction that matches how your audience expects the process to unfold.
Before sharing, review the result for crossings, tiny flows, duplicate labels, and unexplained gaps. If a secondary category adds little insight, consider combining it with similar categories or moving the detail to a table.
Consider testing the chart at the size readers will actually see. A layout that works on a large monitor may be less readable in a report or mobile view. Check whether labels remain readable without hovering.
Interactivity can help when readers need to inspect a surprising path. Filters, hover details, and drill-downs can reveal context without crowding one static view. Learn how interactive data exploration can support that follow-up process.
Available formats depend on your visualization platform and selected chart library. Check the library, data requirements, and maintenance needs before planning a Sankey implementation.
Design Checks Before You Share
Consider these checks before publishing or presenting the chart:
Title: Consider naming the measure, unit, and timeframe.
Labels: Use plain language for nodes and values when space allows.
Direction: Make the intended reading path easy to identify.
Color: Give each color a defined purpose.
Scale: Check that widths follow the stated method.
Clutter: Consider limiting stages, crossings, and minor flows.
Source: Show where the data came from.
Accessibility: Consider a text summary or companion table.
Clear labels, scales, and exceptions can help readers interpret the chart as intended. Before sharing, check whether the visual could suggest more precision than the data supports.
If the chart will appear in a presentation, report, and dashboard, review each version separately. Consider whether small labels, subtle colors, or hover-only details remain usable in every destination. A companion table may help readers verify specific values.
Sankey Diagram vs. Other Flow Visualizations
Related chart types can look similar, and their labels may vary across platforms. Use the comparison below as decision guidance, then check how your selected platform defines each format.
| Visualization | Typical Focus | Encoding to Check | Typical Use |
|---|---|---|---|
| Sankey diagram | Directional movement between connected nodes | Whether link width represents a shared measured value | Consider when volume splits, combines, or ends across stages |
| Funnel chart | Progress through an ordered sequence | Whether section size represents the value at each stage | Consider for a concise conversion or drop-off view |
| Alluvial diagram | A related flow view across categories or stages | Definitions and band meanings can vary by implementation | Consider when your platform defines it for category movement |
| Network graph | Connections among entities or nodes | Line, direction, width, and color meanings can vary | Consider when connections matter more than measured directional volume |
“Alluvial diagram” can refer to a related flow-chart form, but usage is not always consistent. Inspect the labels, direction, width encoding, and platform documentation rather than relying only on the name.
Ask what each line or band represents, whether width carries a value, and whether stages have a required order. Those answers can help you compare formats without assuming a universal taxonomy.
The best option also depends on data volume, refresh frequency, sharing needs, and reader skill. Compare data visualization tool features against those requirements before selecting a platform.
Conclusion
A Sankey diagram can be useful when directional movement and proportional volume sit at the center of your question. Consider a consistent source-target-value table, clear measure and timeframe labels, and a view that keeps major paths easy to follow.
To explore ThoughtSpot, book a ThoughtSpot demo.
