Hierarchy & flow

Sankey diagram

Use a sankey diagram to answer 'where does it go?': flows between stages where the width of each band is proportional to quantity, like traffic source through to purchase. It's the right choice when a flow branches, merges, or leaks at multiple points, which a simple funnel can't represent. Keep the node count modest; too many nodes and links turn the diagram into a tangle.

Minimal dvt spec

The panel body for a sankey diagram. Add an id and title of your own; the sample data is inline, so it renders the moment you paste it into a dvt dashboard. Point it at your warehouse by swapping that sample data for a data block whose query returns the same shape.

View spec

{
  "type": "chart:sankey",
  "data": {
    "rows": [
      {
        "source": "Visit",
        "target": "Signup",
        "value": 60
      },
      {
        "source": "Visit",
        "target": "Churn",
        "value": 40
      },
      {
        "source": "Signup",
        "target": "Trial",
        "value": 38
      }
    ]
  },
  "spec": {
    "sourceField": "source",
    "targetField": "target",
    "valueField": "value",
    "series": [
      {
        "type": "sankey"
      }
    ]
  }
}

Customize

Every control in dvt's edit tray writes a field of this spec; the variants below change the minimal spec above, and the reference below lists every control and ChartSpec field this type accepts.

Palette

Apply the set2 palette so each node takes its own hue.

What changed

 {
   "type": "chart:sankey",
   "data": {
     "rows": [
       {
         "source": "Visit",
         "target": "Signup",
         "value": 60
       },
       {
         "source": "Visit",
         "target": "Churn",
         "value": 40
       },
       {
         "source": "Signup",
         "target": "Trial",
         "value": 38
       }
     ]
   },
   "spec": {
     "sourceField": "source",
     "targetField": "target",
     "valueField": "value",
     "series": [
       {
         "type": "sankey"
       }
-    ]
+    ],
+    "palette": "set2"
   }
 }

Full spec

{
  "type": "chart:sankey",
  "data": {
    "rows": [
      {
        "source": "Visit",
        "target": "Signup",
        "value": 60
      },
      {
        "source": "Visit",
        "target": "Churn",
        "value": 40
      },
      {
        "source": "Signup",
        "target": "Trial",
        "value": 38
      }
    ]
  },
  "spec": {
    "sourceField": "source",
    "targetField": "target",
    "valueField": "value",
    "series": [
      {
        "type": "sankey"
      }
    ],
    "palette": "set2"
  }
}

Everything you can set

Control	Layer	Where it lands
Style
Series colors	Core	`overrides.chart.series.1`
Labels
Show data labels	ECharts	`spec.series[].label.show`

Chart spec fields

Every chart type accepts these ChartSpec fields. Most are dvt spec fields the renderer compiles for you. grid passes straight through to ECharts; series mixes ECharts series options with dvt's dataField; and any key on xAxis, yAxis, legend, or tooltip that is not a documented dvt key passes through too.

Field	What it does
`xAxis`	One axis, or an array of axes, via `AxisSpec`: type, name, min, max, scale, log, and label formatting.
`yAxis`	Same shape as `xAxis`, for the value axis: type, name, min, max, scale, log, and label formatting.
`series`	Per-series overrides: `type`, `dataField`, `name`, `stack`, `smooth`, `itemStyle`, `lineStyle`, `areaStyle`, `label`.
`legend`	Show or hide the legend, set its position, and format an aggregated value shown next to each entry.
`tooltip`	Choose which fields appear, a totals row, field order, a custom template, and crosshair behavior.
`grid`	A raw ECharts grid box (`left`, `right`, `top`, `bottom`, `containLabel`), passed through unmodified.
`funnelRate`	Show a conversion-rate label on a funnel: `step`, `overall`, `total`, or `none`, with a `precision`.
`label`	Data labels: `show`, `position`, `format`, a derived value (percent of total, delta), rotation, size, color.
`gridlines`	Toggle and style gridlines per axis: `show`, dashed or solid `style`, `width`, and `color`.
`banding`	Alternating background bands along one axis, to make long rows or columns easier to scan.
`plotArea`	Set the plot area's background fill and border color.
`gridPadding`	Explicit `left`, `right`, `top`, and `bottom` padding around the plot area, each a number or a string.
`density`	Overall plot density: `comfortable` (default spacing) or `compact` (tighter margins).
`palette`	Name a registered color scheme to apply to the series, instead of the default categorical colors.
`categoryColors`	Map specific category values to explicit colors, overriding the palette for just those categories.
`colorRules`	Apply a color to a data point the first time one of its ordered `when` conditions matches.
`colorScale`	A continuous color scale (`sequential`, `diverging`, or `piecewise`) with a scheme, domain, and buckets.
`annotations`	Fixed or computed reference lines, bands, points, or text, placed at an axis value or a stat of a field.
`trendline`	Overlay a fitted trend line: `true` for a default linear fit, or an object naming the method and order.
`footnotes`	Up to 50 footnote entries, each a marker plus explanatory text, rendered below the chart.
`sourceNote`	A single line of source-attribution text rendered below the chart.

Build it in dvt

Every chart type on this site is a few lines of the same declarative JSON spec. Read the full spec format, or connect an AI agent to your warehouse and build one live in the quickstart.

Follow the build

dvt is in founding research. Leave your email and we'll reach out when there's something to see — no newsletters, no noise.