Showing what a total is made of.

Correct when the reader's question is genuinely about shares of one total, and wrong more often than it is used.

Drawn as a stacked bar, pie chart, treemap, depending on the data.

Where unrestricted income comes from

One household is 14% on a pledge that ends this year

Rendered from the project Donor file — FY2023 to FY2026, through the same compiler the product uses. Not a picture of one.

What it is for

A composition chart answers what proportion of a total each part represents. The test for whether it is the right form is whether the total is meaningful in itself: shares of unrestricted income, of stock value, of a no-show rate broken into its three components. If summing the categories produces a number nobody would quote, this is the wrong form.

Its weakness is comparison. Only the bottom segment of a stacked bar sits on a common baseline; every segment above it starts at a different place in each bar, so comparing the third segment across five bars is a task the chart makes harder rather than easier.

Reach for it when

The total is meaningful and the question is about shares of it
There are few enough parts to label directly, generally under about six
One or two segments are the finding and the rest are context
The parts genuinely sum to the whole with nothing missing or double counted

Use something else when

The reader needs to compare a middle segment across several bars
There are many small categories, which become an unreadable band of slivers
The categories overlap, so a value could belong to two of them

How this form misleads

A stacked bar where the comparison was the point

If the question is how each segment changed between periods, only the bottom one is legible. The reader ends up estimating differences between shifting baselines.

Split it into small multiples, one panel per segment on a common baseline, or plot the segment that is the finding on its own.

A pie chart with too many slices

Angle is read poorly, and beyond about five slices the labels have to go outside the chart, at which point the pie is a legend with a decoration attached.

Sort into a bar chart, or group the tail into a named residual and say how many categories it contains.

Percentages that do not sum to the whole

Where categories overlap or some records fall outside all of them, the parts do not make the total, and a composition chart asserts that they do.

State the base explicitly and show the residual as its own segment rather than letting the chart round the discrepancy away.

What this will not draw, whatever you ask for

The agent picks an intent from a closed list and a single compiler renders it. These forms are rejected at the schema layer rather than discouraged in a style guide, which is the difference between a rule and a preference.

3D anything: Perspective makes two equal values look different. There is no case where the third dimension carries data.
Gauges and speedometers: An enormous amount of ink for one number, and the dial implies a range that is usually invented.
Radar charts: Area scales with the square of the values and the shape changes entirely if you reorder the axes.
Word clouds: Size encodes frequency, which is the least interesting thing about a word, and nothing is comparable.
Pie charts above five slices: Angles are hard to compare, and beyond five slices the labels leave the chart anyway.
Per-series arbitrary colour: Colour means something specific in this product. A series that picks its own hue breaks the meaning everywhere else.

Make one from your own data

You do not choose the form. The analysis works out which question your data can answer and picks the chart that answers it, then tells you why.

No account needed to start. You only pay when you like what you see.