Chartjunk

Ink that shows no data gets in the way.

Heavy gridlines, ornamented axes, textured backgrounds, gratuitous 3-D, and decorative pictures add ink without adding information. Some of it wastes attention; some of it distorts the data itself.

Origin

Edward Tufte coined the term in The Visual Display of Quantitative Information (1983), building on ideas from a Princeton statistics course he taught with John Tukey. Decoration, he argued, generates ink that tells the viewer nothing new, and good graphics devote as much of their ink as possible to the data.

He singled out the illustrated charts of Time’s Nigel Holmes, who replied that Tufte was trapped in the world of academia and blind to the commercial need to grab an audience. Research has since complicated the picture: embellished charts can be remembered better, and practitioners such as Stephen Few and Robert Kosara argue that the definition is too loose.

The Principle

Chartjunk is any visual element in a chart that isn’t needed to understand the data, or that distracts from it. Some of it only competes for attention, like heavy grids and frames; the worse kind distorts the data, like 3-D bars and pictures drawn out of scale.

Origin
Tufte (1983)
Also called
Non-data-ink
In practice
Remove ink that does not show data
When to Use
How to Use
The classic 01 / 10

Ink That Says Nothing

Frames, dark grids, ornamented axes, and shaded backgrounds add ink that tells the viewer nothing new. Strip them away and the same data reads faster and more clearly.

Best for
Explaining the idea
Use when
A chart carries decoration
Avoid when
Every element encodes data
RelatedData Visualization
Gridlines 02 / 10

Lighten the Grid

Heavy or dense gridlines compete with the data for attention. Lighten and thin them, or remove them entirely, and keep only what helps people read values.

Best for
Charts and tables
Use when
Gridlines are dark or dense
Avoid when
Exact values must be read off the grid
RelatedNegative Space
3-D 03 / 10

A Third Dimension Distorts

3-D bars and pies add a dimension that carries no data, and perspective makes equal values look unequal. Flat charts compare honestly.

Best for
Bar, pie, and area charts
Use when
Considering 3-D effects
Avoid when
Depth is used only as decoration
RelatedMüller-Lyer Illusion
Backgrounds 04 / 10

Busy Backgrounds Hide Comparisons

Patterns, photos, and noisy textures behind a chart make bars and lines harder to compare. Put data on a plain background, and keep imagery beside the chart rather than under it.

Best for
Infographics and hero charts
Use when
A chart sits on an image or pattern
Avoid when
The background is already plain
Debate 05 / 10

Tufte Versus Holmes

Tufte attacked Nigel Holmes’s illustrated charts for Time as chartjunk; Holmes answered that commercial graphics need to grab an audience. The tension between statistical rigor and visual appeal hasn’t gone away.

Best for
Editorial and marketing graphics
Use when
Charts must attract attention
Avoid when
Charts support precise decisions
Memory 06 / 10

Useful Junk?

A 2010 study found that Holmes-style embellished charts were read as accurately as plain ones and remembered better weeks later, and a 2013 study found that pictograms make visualizations more memorable. Decoration has a cost, but not always a net one.

Best for
Presentations and infographics
Use when
Recall matters more than precision
Avoid when
Dashboards read daily for exact values
RelatedPicture Superiority Effect
Judgment 07 / 10

Not All Junk Is Equal

Robert Kosara sorts embellishment into harmful, such as a busy background; harmless, such as a tasteful border; and helpful, such as an annotation. Judge each element by what it does to understanding.

Best for
Chart reviews
Use when
Deciding what to remove
Avoid when
Stripping annotations that explain
Accessibility 08 / 10

Icons Can Help

A 2021 study found that semantically meaningful icons made charts more accessible to people with intellectual and developmental disabilities. The most minimal chart isn’t always the most understandable one.

Best for
Charts for broad audiences
Use when
Readers vary in numeracy
Avoid when
Icons carry no meaning
✦

Chartjunk in the Age of AI

Generated charts tend to arrive styled: gradients, shadows, glows, and icons. The data-ink question is now part of reviewing AI output.

✦ AI Era 09 / 10

Generated Charts Love Decoration

AI chart tools often add gradients, rounded 3-D bars, shadows, and stock icons by default. Ask for plain encodings, and review the result as you would a designer’s draft.

Shift
Default styling → plain encoding
Use when
Generating charts with AI
Watch for
Accepting decorative defaults
✦ AI Era 10 / 10

Annotations Are Not Junk

AI can write short annotations that point out what matters in a chart, the helpful kind of extra ink. Check every number it states against the data.

Shift
Decoration → explanation
Use when
AI annotates charts
Watch for
Annotations with unverified numbers
Further Reading

Research

Practice

Wikipedia