This entry is part of the Chart Index, the reference library for the Chart Design Field Guide.
The slope chart is a line chart with exactly two time points. That constraint is its strength. With only a before and after, the eye reads the line as a single piece of information: direction, magnitude, and rank change, all in one glyph. The chart trades the granularity of a multi-point line for a clarity of comparison that no other form offers.
Tufte calls this a small noncomparative, highly labelled display. Each line in a slope chart is a single category compared between two states, and the slope itself — flat, climbing, falling, crossing — is the message.
What it is
A slope chart maps two time points (or two states) to the left and right edges of the chart, and a quantitative variable to vertical position. Each category is rendered as a line connecting its starting value (left) to its ending value (right). The slope of each line encodes the rate of change; the crossing of lines encodes rank changes.
Six segments, two states. Four segments climbed; one stayed flat; one fell. The crossing lines (Education dropping below the climbers) is the story — and it is visible without reading a single value. A bar chart of the same data would have hidden the direction of change behind two parallel rows of bars.
When to use it
Slope charts are the right choice when:
- You have exactly two time points (or two comparable states): before/after, pre/post, last year / this year.
- The reader's question is "who moved up, who moved down, by how much?"
- You have 5–15 categories — enough that ranking matters, few enough that the lines are distinguishable.
- The direction of change matters more than absolute values.
- You want to show rank crossings — which lines pass which.
When not to use it
- Three or more time points. A slope chart with three points is just a line chart. Use a line chart.
- Many categories. Twenty crossing lines becomes a tangled web. Use small multiples or a heatmap of changes.
- Comparing magnitudes precisely. Slope encodes change, not absolute value. For magnitude comparison, a bar chart is more accurate.
- No category-level identity. If the categories are not stable (different categories pre and post), the connecting line is misleading.
Design principles
Direct-label every endpoint
Every line ends on the right; every label sits next to its endpoint. Reading order: eye lands on right endpoint, picks up the label and value, traces back to the left endpoint. No legend, no axis, no chartjunk.
Use consistent axis range across both ends
Both the left and right edges share the same value scale. The slope is meaningful only if the y-axis is uniform across the chart. A common error: scaling each side independently, which fakes the changes.
Highlight the lines with a story
If one or two lines carry the message (a segment that dropped, a segment that crossed), give them the accent colour and weight. Other lines fade to a neutral grey. The hero lines pull the eye; the context lines provide the rank baseline.
Strip the y-axis
Tufte's slope charts have no y-axis. The endpoints are labelled with their values, the slope is the encoding, and an axis would only add ink without adding information. Let the labels carry the precision.
Anchor at zero only if proportional change matters
Like a line chart, the y-axis of a slope chart usually starts above zero — the data range, padded modestly. The exception is when proportional change is the message; then anchor at zero.
Show small multiples for many categories
If you have twenty categories, do not put them all on one slope chart. Break into themed small multiples — one panel per region, per segment, per cohort — and let each panel hold a manageable number of slopes.
Annotate the rank crossings
When two lines cross, that crossing is often the story. Mark it with a small note, or use line weight to draw attention. Education fell below Enterprise post-launch — a sentence-long annotation explains what the chart shows.
Anatomy
The slope chart is structurally minimal: two vertical guides, a set of lines with endpoint dots, and direct labels at each end. Tufte's preferred form for two-point comparison.
Related types
- Line chart — for three or more time points. Slope charts are a degenerate case.
- Grouped bar — the conventional alternative. Encodes the same data but hides direction.
- Dumbbell plot — two dots per row joined by a line. Like a horizontal slope chart, useful for many categories.
- Bump chart — slope chart for ranks, with multiple time points.
- Connected scatter — points in 2D space connected in order. Same idea, two variables instead of one over time.
Reading list
- Tufte, E. (1983). The Visual Display of Quantitative Information. The original treatment of the slope graph.
- Few, S. (2012). Show Me the Numbers. The case for slope charts in business communication.
- Cleveland, W. & McGill, R. (1984). Graphical Perception. Why slope is read pre-cognitively.