This entry is part of the Chart Index, the reference library for the Chart Design Field Guide.

An area bump chart shows how both rank and magnitude change over time. Each category is a flowing ribbon: its vertical position at each step is its rank, and its thickness is its value. It is the answer to the bump chart's one weakness — that a line at rank 2 looks the same whether it is a whisker behind the leader or half its size — and to the stacked area's — that a layer buried in the middle of a stack cannot be compared with one at the bottom.

The form is younger than either parent. It grew out of the streamgraph work of the late 2000s and the ranked-flow diagrams that newsrooms began using for market share, genre popularity and sporting ladders: charts where the story is who is on top now, and by how much, rather than how big is the whole. It is an editorial chart. It rewards a small number of categories with a clear story and punishes everything else.

What it is

An area bump chart maps time to the x-axis and, at each time step, sorts the categories by value and stacks them in that order — largest at the top, smallest at the bottom — with a small gap between them. Each category's ribbon is drawn through its slot at every step, so a ribbon that rises has overtaken the ones it crosses, and a ribbon that thickens has grown. The vertical position of any edge means nothing on its own; only order and thickness carry data. Labels sit at both ends of each ribbon.

The hero chart above is six years of one company's media budget, by channel, as share of spend. Television starts as the widest ribbon and the top rank; by FY26 it is less than half the width and in third place. Social does the opposite, thickening every year and climbing from fourth to first. Print thins to almost nothing and sinks to the bottom. The bump-chart reading — who overtook whom, and when — is all there in the crossings; the extra reading, that Social's rise was nearly a tripling and Print's fall was to less than a third, is in the ribbon widths. Randomise it and each year's shares jitter, the ranks are recomputed, and the ribbons re-route.

When to use it

Area bump charts are the right choice when:

  • Rank changes over time matter, but equal-width bump lines would hide large differences in value.
  • A small number of categories — four to eight — compete for share, volume, or attention.
  • The reader needs to see both who moved and whether the movement was commercially meaningful.
  • The x-axis has few steps — four to ten — so the ribbons have room to run flat between them.
  • The audience will read it editorially, for a story, rather than to look up numbers.

When not to use it

  • Precise values are the primary task. A ribbon's thickness can be compared, not read. Use a line or grouped bar chart.
  • Many categories or frequent rank changes. Beyond eight ribbons, or with ranks that reshuffle at every step, the crossings become a knot.
  • Values can be negative. Band thickness has no natural representation for a negative quantity.
  • The total matters. The ribbons are separated by gaps and re-sorted at each step, so the stack's height is not the total. That is a stacked area's job.
  • Time steps are many. With sixty points the flat runs vanish and the chart degrades into a streamgraph with worse ordering.
The hero's channels as a plain bump chart
The same six years with thickness removed. Every overtaking is still here, and the chart is cleaner for it — but Social's climb from fourth to first and Email's from seventh to second now look like the same event, and nothing says that one of them is a quarter of the budget and the other an eighth.

Design principles

Limit the number of bands

Five to eight series is usually enough. Beyond that, crossings and narrow bands compete for attention, and the thin ribbons at the bottom — the ones the form is worst at — multiply.

Fourteen product categories
Fourteen ribbons over six quarters. The top three can still be followed; the bottom eight are threads a few pixels wide that cross each other at every step, and the end labels have to be nudged apart to fit. Aggregate the small ones into an Other, or cut to the top six.

Label bands directly

Place category names at the start or end of each band — the house renderer prints both. A detached legend turns tracing into a memory task: the reader looks up a colour, finds the ribbon, follows it, and has forgotten the name by the time they reach the other end.

Keep colour stable

A category must retain the same colour as it changes rank. Colour identifies the path; position reports its rank. Colouring by rank instead — gold for first, silver for second — makes every ribbon change hue at every crossing and destroys the one thing the reader was using to follow it.

Explain thickness

State the unit represented by band thickness. Without that cue, readers may treat the chart as a decorative bump chart. Ribbon thickness is share of spend in the hero's subtitle is the whole difference between a chart and an illustration.

Give the ribbons room to be read

Two spacing decisions shape the form. The gap between ribbons is what separates an area bump from a stacked area: close it and the ribbons fuse into a single stack that the eye reads as a total. The flat run at each step is what lets a thickness be compared: shrink it and every ribbon is always in transit.

No gap between ribbons
With the spacing at zero the ribbons touch, and the chart reads as a stacked area whose layers keep swapping places. The eye tries to read the top edge as a total, which it is not, and loses the individual thicknesses in the mass.
A short flat run at each step
With the flat run cut to a fifth of the step, the ribbons are almost always crossing and there is nowhere to compare two thicknesses side by side. The default run of just over half the step is what makes each year readable as a column of widths.
Default spacing and run
The same five channels with the house defaults: an eight-pixel gap and a flat run of sixty per cent of each step. Each year is a column of separated widths that can be compared top to bottom, and the crossings between columns are short and legible.

Use few time steps

The form's flat runs need width. Six annual steps give each ribbon a readable plateau; sixty monthly steps turn the plateaus into stubs and the chart into a nervous streamgraph. If the data is monthly, aggregate to quarters or years, or switch to a streamgraph and give up the rank ordering.

Twenty-four monthly steps
The same five channels at monthly resolution. There are now twenty-three sets of crossings, the flat runs are a few pixels wide, and the eye reads a wobble rather than a sequence of standings. Aggregate to years, as the hero does, and the story returns.

Put the story at the top

The renderer sorts largest to the top at every step, so the chart's most prominent region — the top edge — is always the current leader. Order the underlying series so the categories the prose discusses take the first colours in the palette, and keep the small, stable categories in the quiet greys so the eye is not drawn to the bottom of the chart where nothing is happening.

Anatomy

The composition of an area bump chart
An anatomical guide

The x-axis is ordered time, the vertical order is rank, band thickness is the quantitative value, and crossings mark changes in rank. The gaps between ribbons are what keep it from being read as a stack.

  • Bump chart — rank change without magnitude.
  • Streamgraph and stacked area — composition over time without rank ordering, and with the total intact.
  • Slope chart — two time points, with values, for the before-and-after version of the same question.
  • Line chart — precise change in values over time.
  • Marimekko — the other chart that encodes two quantities in width and height at once, for a single period.

Reading list

  • Cleveland, W. S. (1985). The Elements of Graphing Data. Foundations for reading position and change.
  • Byron, L. & Wattenberg, M. (2008). Stacked Graphs — Geometry & Aesthetics. The layered-ribbon geometry the area bump borrows.
  • Wilke, C. (2019). Fundamentals of Data Visualization. On stacked and ranked displays of proportions over time.