Data Visualization

Charts That Actually Communicate

Sixteen chart types grouped by the question they answer — comparison, relationship, distribution, and composition. Pick the chart that fits the question, not the one that looks impressive.

Comparison
Ranking

Bar Chart

Horizontal bars excel at ranking. Long category names read naturally left-to-right, and dozens of items stack tidily down the page without sacrificing label clarity.

Top product categories ApparelHomeToysSportsBooksAuto
Best for
Ranking · Top-N lists
Use when
Long labels · Many items
Avoid when
Time-series data
Comparison

Column Chart

The workhorse of comparison. Vertical columns let the eye judge length along a common baseline — one of the most accurate visual encodings for categorical data with short labels.

Revenue by region NAEUAPACLATAMMEAF
Best for
Comparing categories · Time-series
Use when
5–12 items · Short labels
Avoid when
Labels are long or angled
Trend

Line Chart

The clearest way to show change over a continuous variable — usually time. Slope tells you direction, steepness tells you rate, crossings tell you where one series overtakes another.

Monthly active users JanAprJulOct
Best for
Trends over time
Use when
Continuous data · Many points
Avoid when
Categories aren't sequential
Comparison

Radar Chart

Plots several quantitative variables on spokes radiating from a shared center. Best for comparing the overall "shape" of a few items across the same set of metrics.

Skill profile SpeedPowerRangeCostUX
Best for
Multi-metric profiles
Use when
3–8 axes · 1–2 series
Avoid when
Many overlapping series
Relationship
Relationship

Scatter Plot

Each point is one observation across two variables. Reveals correlation, clusters, and outliers — the chart that lets the data's structure show itself, rather than imposing one.

Hours vs. score Hours studied Score
Best for
Correlation · Outliers
Use when
Two numeric variables
Avoid when
Tracking change over time
Relationship

Bubble Chart

A scatter plot with a third variable encoded as dot size. Reveals relationships across three dimensions at once — x position, y position, and magnitude.

Revenue · users · margin Revenue Users
Best for
Three numeric variables
Use when
Size adds real meaning
Avoid when
Bubbles overlap heavily
Relationship

Connected Scatter

A scatter plot whose points are joined in time order. Shows how the relationship between two variables evolves — tracing a path rather than a static cloud.

Inflation vs. rates Inflation Rate
Best for
Two variables over time
Use when
Order / trajectory matters
Avoid when
No natural ordering
Density

Heatmap

A grid where color intensity encodes a third value across two dimensions. Excellent for spotting density and pattern at a glance — calendar usage, correlation matrices, geographic clusters.

Active users · day × hour MonTueWedThuFriSatSun 0h6h12h18h
Best for
Density across two dims
Use when
Patterns matter more than exact values
Avoid when
Accuracy is critical
Distribution
Distribution

Histogram

A bar chart for a single continuous variable, binned into ranges. Reveals the shape of a distribution — where the mass is, whether it's skewed, whether you're looking at one population or several.

Response time (ms) Response time (ms)
Best for
Distribution shape
Use when
One continuous variable
Avoid when
Comparing categories
Statistical

Box Plot

A compact summary of a distribution: median, quartiles, range, and outliers — all in one mark. Indispensable when comparing the spread of many groups side-by-side.

Response time by region NAEUAPACSA
Best for
Comparing distributions
Use when
Spread + median matter
Avoid when
Audience isn't stats-literate
Distribution

Violin Plot

A box plot crossed with a density curve — the width at each value shows where data clusters. Reveals multi-modal shapes a plain box plot would hide.

Score distribution Cohort ACohort BCohort C
Best for
Comparing full distributions
Use when
Shape matters, not just quartiles
Avoid when
Audience isn't stats-literate
Distribution

Strip Plot

Every observation plotted along one axis, jittered to avoid overlap. Honest about sample size and outliers — you see the real data points, not a summary of them.

Response time by tier FreeProEnterprise
Best for
Small-to-mid datasets
Use when
Showing every observation
Avoid when
Thousands of points overlap
Composition
Proportion

Pie & Donut

Parts of a whole, expressed as angle. Reliable only for two or three slices — beyond that, eyes can't distinguish small angular differences and a bar chart will read clearer.

Browser share
Chrome50%
Safari30%
Other20%
Best for
Two-way splits
Use when
≤ 3 slices · Whole = 100%
Avoid when
Many slices · Close values
Volume

Area Chart

A line chart with the area below filled in. Adds visual weight that emphasizes volume or cumulative magnitude — useful for showing how a total accumulates or how parts stack up.

Hours logged JanAprJulOct
Best for
Volume + trend together
Use when
Cumulative or stacked totals
Avoid when
Series overlap and hide each other
Hierarchy

Treemap

Hierarchical proportions packed into a rectangle. Each block's area represents its value, and nesting shows containment — efficient for showing both share and structure in a single dense view.

Budget allocation Engineering42% Sales Marketing Design Ops Other
Best for
Hierarchical part-of-whole
Use when
Many categories · Nested data
Avoid when
Precise comparison is needed
Flow

Sankey Diagram

Shows flow between states. Thickness of each ribbon represents quantity transferred — the best chart for tracing how a user, a dollar, or an electron moves through a system.

Signup → activation flow Signups ActivatedEngagedChurned
Best for
Flow between states
Use when
Funnels · Energy · Network flow
Avoid when
Many cross-overs make it tangled

Data Viz in the Age of AI

Charts are increasingly generated from a sentence — and the model can pick the encoding and narrate the insight for you.

✦ AI Era

Describe, Don’t Configure

Natural-language charting: "show revenue by region as bars" produces the chart. The prompt replaces the chart-type picker and the axis menus.

Shift
Configure → describe
Use when
Exploratory analysis
Watch for
Misread intent → wrong chart
✦ AI Era

AI Picks the Encoding

Given raw data, a model can suggest the right chart type and narrate the takeaway in plain language above it — a first draft a human still edits.

Adds
Suggested type + narration
Use when
User isn’t a chart expert
Watch for
Confident wrong insight
Further Reading