Area Chart Maker

Takes the same wide CSV as the line chart and stacks the series so each band sits on top of the one below it. The top edge is the total and each band is a component contribution. That stacking is the whole point, and it is also the source of every difficulty with the format.

How to use it

1. Paste wide CSV: first column categories, remaining columns the components to stack.
2. Order the columns so the largest and most stable component sits at the bottom.
3. Edit values and watch the bands redraw.
4. Choose a theme and export a high-resolution PNG.

What stacking actually shows

In a stacked area chart, only the bottom series is measured from zero. Every other band is measured from the top of the one beneath it, so its baseline moves as the data changes.

This gives you two readings for free. The top edge of the stack is the total across all components, which is often the most important line on the chart. And the thickness of each band is that component contribution at each point.

What it makes hard is reading the shape of any series other than the bottom one. Because the baseline underneath is moving, a band with constant thickness looks like it is rising and falling. The human eye is poor at judging thickness between two curved edges — it tends to read the perpendicular distance rather than the vertical distance, which understates thickness on steep slopes. So a component that is genuinely flat can look volatile, and a genuinely volatile one can look smooth.

The practical consequence is a rule about ordering: put the largest and most stable series at the bottom, where it has a flat baseline and distorts the least, and the smaller or noisier ones above it. If two orderings both seem defensible, the one that produces the flattest lower bands is the one that will mislead least.

When to use it and when not to

A stacked area chart is the right choice when the total is meaningful and the components sum to it. Revenue by product line, traffic by channel, and cost by department all qualify — the sum is a real quantity someone cares about, and the parts genuinely make up the whole.

It is the wrong choice when the series are independent. Stacking three unrelated metrics produces a top edge that is the sum of things that should not be added, which reads as a meaningful total and is not one. Use a line chart for independent series.

It is also wrong when the reader needs to compare individual series precisely. If the question is whether channel B grew faster than channel C, the moving baselines make that comparison unreliable, and an unstacked line chart or small multiples will answer it directly.

Negative values break stacking entirely. A stack assumes components accumulate in one direction, and a negative component would need to extend below its baseline, which produces overlapping bands that cannot be read. If your data has negatives, a grouped bar chart or a waterfall is the appropriate format.

The 100 percent variant and what it trades away

A common alternative normalises each column so the stack always fills the full height, showing share rather than absolute quantity. It is worth knowing what that swap costs.

Normalising makes composition changes very easy to see. If one channel goes from 40 to 55 percent of traffic, that is unmistakable on a normalised chart and can be nearly invisible on an absolute one if the total is also growing.

What it discards is the total, and the total is frequently the more important fact. A channel holding a steady 30 percent share looks flat, whether the total tripled or halved. Presented alone, a normalised chart can make a collapsing business look stable, which is why it is usually shown next to a total rather than in place of one.

Opacity, colour, and legibility

Area fills are rendered with some transparency so that gridlines and overlapping edges remain visible. This helps when bands are thin and hurts when it makes adjacent colours harder to tell apart, since a semi-transparent fill shifts toward the background colour.

Keep the band count low. Above five or six components the thin bands become unreadable regardless of colour choice, and the honest fix is to group the small ones into an "Other" band rather than to keep adding hues.

Where the components have a natural order — small to large, or a sequence of stages — a sequential colour ramp reads better than a categorical palette, because it reinforces the ordering rather than fighting it. Where they are unordered categories, a categorical palette with distinguishable lightness values is safer than one that varies only in hue, for the same colour-vision reasons that apply to line charts.

At a glance

Frequently asked questions

Read more

Charting change over time — Connecting two points asserts that the values between them exist. That assertion is the whole difference between a line chart and a bar chart.

Related chart makers