Good Charts

Key Ideas

1. Charts are arguments

Every chart asks the audience to notice something. If the argument is unclear, the visual cannot rescue the message.

A chart should be evaluated by the claim it helps the audience inspect. If the audience cannot tell whether the chart is about growth, decline, outliers, ranking, or uncertainty, the visual is not communicating yet. A good chart makes the intended comparison easier than reading the raw table.

Why it matters: a chart without an argument makes the audience do too much interpretive labor. Apply this by asking one reviewer what the chart says before you explain it. If their answer differs from yours, the chart needs stronger visual hierarchy.

For a ranking chart, ask whether the order itself tells the story. If a sales-region chart is alphabetical, the audience must work to see the leaders and laggards. Sorting by value and labeling the top exception may do more for communication than adding color. Good Charts trains that kind of practical visual judgment.

A sales leaderboard, for instance, may need direct labels on the top and bottom three bars while the middle stays quiet. The chart then supports a coaching or resourcing discussion instead of forcing the audience to inspect every value with equal effort.

The audience should be able to tell what is unusually high, low, changed, or missing without waiting for the presenter to narrate every bar.

2. Exploration and explanation differ

A chart used to discover patterns can be messier than a chart used to persuade. Do not present exploratory clutter as if it were a finished explanation.

Exploratory charts can contain extra cuts, labels, and variations because the analyst is still looking for patterns. Explanatory charts need stronger hierarchy. Before presenting, the communicator should remove exploratory residue and rebuild the chart around the point the audience must understand.

Why it matters: exploratory clutter often looks like rigor but feels like confusion in a meeting. Apply this by creating two versions: the messy analysis view for yourself and a presentation view that preserves only the comparison the audience needs.

An exploratory scatterplot with twenty labels may help an analyst discover clusters, but it will overwhelm a leadership slide. The presentation version may need only the outliers, a trend line, and a headline explaining the implication. This distinction keeps analysis artifacts from being mistaken for audience-ready communication.

A common fix is to move dense exploratory visuals to an appendix and rebuild the meeting slide as one clean comparison. Good Charts is useful because it legitimizes that edit: clarity for the audience is not a betrayal of the underlying analysis.

This is especially important when analysts export directly from BI tools; the default chart often preserves filters and labels useful only to the maker.

3. Form should follow comparison

The best chart type depends on the relationship: ranking, trend, distribution, composition, or correlation. Start with the comparison before choosing the form.

Choosing a form starts with the relationship in the data. A time-series question usually needs a line. A ranking question may need ordered bars. A distribution question may need a histogram or box plot. A composition question may need caution because part-to-whole charts are often misread when there are too many segments.

Why it matters: chart type is a reasoning choice. Apply it by naming the relationship first: trend, rank, distribution, relationship, or composition. Then pick the visual form that makes that relationship easiest to judge without verbal rescue.

Chart choice should start with a sentence such as 'we need to compare adoption over time' or 'we need to see whether deal size relates to sales cycle.' That sentence points to a line chart or scatterplot before software defaults get involved. The reader can use the book as a guardrail against choosing visuals by habit.

If the decision is about whether churn is accelerating, a line chart with a marked intervention date will usually beat a pie chart or table. The reader should be able to explain the visual choice in one sentence before defending the chart's design.

A quick sketch before software can prevent this mistake because it forces the maker to choose the comparison before choosing the tool.

4. Simplification is a communication choice

Removing noise is not dumbing down the data. It is deciding what the audience must see first to understand the point.

Simplification is not the same as distortion. The communicator should remove gridlines, labels, colors, or categories that do not help the audience answer the question. But they should not hide inconvenient comparison points or manipulate scales to intensify the story.

Why it matters: simplification protects attention, but only if it remains honest. Apply this by removing decorative elements before touching the data scale. The viewer should see the signal faster without being pushed toward an exaggerated interpretation.

Simplification has to preserve the honest comparison. Removing gridlines can help; truncating an axis to dramatize a small difference may mislead. A Good Charts reader should review every simplification by asking whether a skeptical but fair audience would still agree with the visual's implied message.

A simplified chart should still invite fair disagreement. If removing a category changes the apparent conclusion, the category may need to stay. If removing heavy gridlines makes the conclusion easier to see without changing it, the edit serves communication.

When in doubt, show the chart to someone outside the project and ask what decision they think it supports.

5. Visual literacy is learned

Most teams can improve chart conversations by sketching alternatives, critiquing examples, and naming what each visual makes easy or hard to see.

Why it matters: visual literacy lets a team improve charts without turning every review into taste. Apply it by asking three questions: what is the chart for, what does the eye see first, and what comparison is harder than it should be?

The book is valuable at the team level because it gives people a shared way to critique visuals. Instead of saying 'I like it' or 'make it prettier,' a team can ask what the chart is for, what comparison it supports, and where the audience's eye goes first.

Teams can use the book in chart critique sessions. Instead of debating taste, ask each reviewer to write the first thing they notice, the comparison they think matters, and the decision the chart supports. If the answers diverge, the chart needs clearer hierarchy or a different form.

This team habit also improves future briefs. Once people share a vocabulary for trend, rank, relationship, and distribution, chart feedback becomes faster. Reviewers can say 'this is a ranking question' rather than 'something feels off.'

Over time, this makes chart review less personal: the team critiques the communication job, not the chart maker's taste.