Sankey Diagram FAQ

A Sankey diagram shows quantities moving between categories or stages. The width of each band represents its value. Use it to explain a budget, energy or material flows, a sales breakdown, or routes through a process. Use one common unit throughout the diagram and check the underlying data before interpreting the picture.

Sankey Diagram is a Windows desktop app for Windows 10 and Windows 11. Check Microsoft Store for current system requirements, price and license terms. The interface and built-in help are in English.

Start with Source, Target and Value columns in Excel. Choose Import Excel / CSV, review the worksheet and mapping, preview the data and import. Then choose Update diagram. You can also paste a table, add rows manually or open an example.

Watch the first-diagram tutorial and use the supplied practice files.

Import XLSX, XLS, CSV, TSV or TXT. For Excel, choose the worksheet and range; for text files, check the separator, encoding and number format. To paste, copy a table in Excel and choose Paste from Excel outside an active cell editor. Pasted data has no source file to remember for later reloads.

Connections uses one row per Source → Target link, with a numeric Value and optional Note. Choose it when your table already lists the connections.

Records across stages uses one row per complete route, such as Customer → Product → Region. Choose two to six ordered stage columns, then either sum a numeric column or count the records. All original columns remain available in the table.

Set First row contains headers to match the selected range. Map the fields to the correct columns and choose the number format explicitly. For example, a comma can be a decimal separator or a thousands separator. Keep units out of numeric cells: enter 60, then set kWh under Diagram Options.

After changing import settings, choose Preview data again. Review the validation messages before importing.

Values must be finite, non-negative numbers. Zero stays in the table but produces no visible band. Missing names, negative values and invalid numbers must be corrected; they are not silently converted to zero. Blank rows are skipped and reported. The preview validates all selected rows even when only a sample is displayed.

No. Duplicate rows remain separate records and keep their notes. The diagram combines the widths of matching connections. If a band looks wider than expected, inspect all contributing rows. Selecting a connection helps you find them in the table.

In Records across stages, Sum of values uses each record’s numeric value as its weight. Count rows gives each record weight one. In the tutorial, five records have an input sum of 120 units or an input count of five. Switching to Count rows clears the old value unit unless you explicitly keep it.

Watch Sum versus Count in the app.

By default, every selected stage is required. You can instead represent missing values as explicit (Missing) nodes. The app does not silently skip a stage. The same category name in different stages represents different nodes; the stage name is included in the caption.

Import copies data into your working table. Enable Remember this file for future reloads if you want to read that file again later. Reload from source reads the saved file when you request it and lets you review the replacement. Save project stores your editable .sad project.

Changes made in Sankey are not written back to Excel or CSV. There is no two-way or automatic synchronization.

Choose Import Excel / CSV, enable Remember this file for future reloads, and check the worksheet, data layout and mapping. Choose Import and remember source; this replaces the current table. Save the project. After updating and saving the source file, choose Data source → Reload from source.

Follow the recurring-report tutorial.

Reload replaces the whole table with the reviewed source data. It does not merge your local edits, renamed categories or added rows. The confirmation explains the replacement. Title, units, labels, palette and custom colors for matching nodes are retained; the filter, sorting and selection are cleared. Undo restores the previous data and view.

Confirming Replace with source data also redraws the diagram. Another Update click is only needed after a further change that makes the diagram stale.

Only saved file contents are read. Save the workbook in Excel and close it before reloading, especially if the file is locked. Formula values come from the workbook’s stored results; Sankey does not recalculate formulas. Update and save the workbook in Excel first if those results are stale.

You can still open the saved .sad project and work with its stored data. A missing file or invalid reload preview leaves the current table unchanged. If the file, worksheet or column layout changes, import it again and explicitly remember the new source. Use Disconnect source to keep the current data without the connection.

Keep a project and its nearby source file together when moving a folder. Remembered sources are local files; network URLs and UNC paths are not supported.

Rendering is manual so you can finish several table edits before redrawing. After editing, importing, pasting, filtering or Undo/Redo, choose Update diagram when the message appears. If you remove every row, update once to clear the previous picture. Export diagram and Copy diagram refresh the result automatically.

The same quantity can appear in more than one transition. A flow of 60 entering an intermediate node and then leaving as 45 and 15 contributes 120 when those three bands are added, but still represents only 60 moving through that node. Check the original input and the final destinations, or read the original-record total in Records across stages.

Node labels show incoming and outgoing quantities for the current view. The app does not automatically balance a model; add explicit loss, reserve or waste flows when appropriate.

The diagram, node values, tooltips and export use the included data. In Connections, filters select individual links. In Records across stages, they select original records and keep every transition of each included route together. Check the Filtered data indicator. Choose All data to clear filtering and sorting.

Yes. Open Diagram Options to set the title, label content, number format, unit, fonts, palette and layout. Choose Name and values to show quantities with node names. Select a node to apply a custom color or rename it throughout the data. These changes support Undo. Update a stale diagram before using node actions.

Use Save project to keep editable data, notes and presentation settings in a .sad file. Use Export diagram for PNG, SVG or PDF to share the finished picture. Export table separately writes the visible table to XLSX, XLS, CSV or TXT.

Saving a project or exporting a table does not overwrite a connected source workbook.

Choose Export diagram → PNG and the Presentation 16:9 preset for a 1920 × 1080 image. Use SVG for vector shapes and text, or PDF for a printable document. PNG can have a transparent background. Copy diagram places a presentation-size image with a white background on the clipboard.

For a dense graph, use a useful filter or larger export size, then inspect the labels at the intended reading size.

Undo and Redo reverse document operations outside an active cell editor. Saving does not clear the history; starting a new edit after Undo clears Redo. Saves retain the previous project version as a .bak file.

The app also keeps periodic recovery copies of unsaved work. Available copies are offered at a later launch and do not overwrite the original project. Save a recovered document to keep it. Recovery is periodic, so the final changes before an interruption may be absent.

Your .sad project and imported working data are stored in files on your PC. Remembering an Excel or CSV source does not upload that workbook or enable two-way synchronization. The app also uses Microsoft Store services and diagnostic reporting. See the Sankey Diagram privacy policy for details.

Open Help in the app, watch the tutorials, or contact DefinedBytes support. Describe the steps, file type and expected result. If an example is needed, use a small copy with fictional or anonymized data.