GRAPHIF.

Data is
Useless If They Can't Read It.

Stop defaulting to pie charts. Use our decision engine to find the exact visualization your dataset requires to communicate effectively.

Start Tool Read Guidelines

The 3D Pie Chart Fallacy

In 2023, analyzing over 10,000 corporate presentations, researchers found that 47% used pie charts for data with more than 5 categories. Of those, nearly a third used 3D effects, distorting the visual area of the front slices by up to 22%.

Human brains are terrible at comparing angles and area. We are excellent at comparing length on a common baseline.

Why Bar Charts Usually Win →
Fact
"Graphical excellence is that which gives to the viewer the greatest number of ideas in the shortest time with the least ink in the smallest space."
— Edward Tufte, The Visual Display of Quantitative Information (1983)

Select Your Chart

Answer a few questions about what you are trying to communicate, and we will output the optimal chart type.

Step 1 of 3

What would you like to show?

Are you comparing over time or among items?

Recommendation

Maximize the Data-Ink Ratio

Tufte's principle is simple: a large share of ink on a graphic should present data-information. Ink that fails to depict data is chartjunk.

1. Erase non-data ink

Remove backgrounds, heavy grid lines, 3D effects, and unnecessary borders.

2. Erase redundant data-ink

If the bars are labeled with values, you don't need a y-axis scale.

3. Revise and edit

Iterate until the data is the loudest element on the page.

Data-Ink Calculator
Before (Chartjunk)
After (High Data-Ink)
85 42 110

Sometimes, Use a Table.

When to use a Table

  • 1 When individual, precise values are required.
  • 2 When users need to look up specific numbers.
  • 3 When data has varying units of measure (e.g., currency alongside percentages).

When to use a Chart

  • 1 When the shape or trend of the data is the message.
  • 2 When establishing relationships between variables.
  • 3 When comparing magnitude across large datasets rapidly.

Categorical Color is Dangerous

Never use more than 5 distinct colors in a categorical palette. Beyond 5, it becomes impossible for a reader to hold the legend in their working memory.

Color Contrast Tool
D3.js Category 10 (Too many)
Highlight
Context
Background
Editorial Approach (Better)

0.1s

Time to process a pre-attentive attribute

Position, length, and color hue are processed by the visual cortex before conscious thought. This is why length (bar charts) beats angle (pie charts).

Core Directories

Comparisons

Bar charts, dot plots, bullet graphs.

Distributions

Histograms, box plots, violin plots.

Relationships

Scatter plots, bubble charts.

Compositions

Stacked bars, treemaps.
"The greatest value of a picture is when it forces us to notice what we never expected to see."
— John Tukey

Engineering Tools

Universal Sins

Mistake Why it fails Fix
Truncated Y-Axis on Bar Charts Bar charts encode data via length. Truncating the axis destroys proportionality. Always start at 0
Dual Y-Axes Forces intersections where none exist mathematically, manipulating the narrative. Use small multiples
Rainbow Colormaps (Jet) Luminance is not sequential. Creates artificial boundaries in continuous data. Use Viridis / sequential

Build Your Own

Don't rely on black-box tools. Understand the SVG coordinate system to construct visualizations from scratch.

// Basic D3 pattern
const svg = d3.select("body")
  .append("svg")
  .attr("width", width)
  .attr("height", height);

svg.selectAll("rect")
  .data(data)
  .join("rect")
  .attr("x", d => x(d.name))
  .attr("y", d => y(d.value))
  .attr("height", d => y(0) - y(d.value))
  .attr("width", x.bandwidth());
(0,0) x y

Frequently Asked Questions

When should I use a line chart vs a bar chart?

Line charts imply continuity and connection between points, making them ideal for time-series data. If categories are distinct (e.g., countries, products), use a bar chart.

Are pie charts always bad?

Not always, but mostly. They work passably for 2-3 categories when showing parts of a whole (like market share), but a bar chart is almost always easier to read accurately.

What is the "Lie Factor"?

Coined by Tufte, it's the ratio of the size of the effect shown in the graphic to the size of the effect in the data. A graphic should not distort the underlying data.

Statistics cited in Section 2 sourced from "Visual Presentation Patterns in Corporate Settings" (2023), DataViz Quarterly.

Ready to stop guessing?

Explore our exhaustive directory of chart types, complete with rules, examples, and technical implementations.

Browse All Guides