Data-Ink Ratio

The data-ink ratio is Edward Tufte's measure of chart efficiency: the proportion of a chart's pixels that display actual data, versus pixels spent on borders, grid lines, and decoration. The goal is to push that ratio as close to 1.0 as possible.

What is the data-ink ratio?

It's the fraction of a chart's total ink (or pixels) devoted to showing actual data, rather than borders, backgrounds, and other non-data decoration — Tufte's shorthand for "how much of this chart is doing real work."

The Formula

Data-Ink Ratio = Data-Ink / Total Ink Used

Goal: Ratio as close to 1.0 as possible

Data-Ink = Pixels showing actual data
Total Ink = All pixels in the chart

High ratio = Clean, focused chart
Low ratio = Cluttered chart with unnecessary elements

Tufte's Principle

"Erase non-data ink, within reason... Erase redundant data-ink, within reason."

Translation: If it doesn't represent data, question whether you need it. If it represents data multiple times, you probably need it only once.

What counts as "non-data ink" on a chart?

Anything that isn't a bar, line, point, or a label needed to read the data — grid lines, borders, background fills, 3D effects, and decorative icons are the usual suspects.

Common Offenders

Element	Is It Data-Ink?	Action
Bars, lines, points	✅ Yes	Keep (this is your data)
Axis labels	✅ Yes	Keep (necessary context)
Heavy grid lines	❌ No	Lighten or remove
Chart borders	❌ No	Remove
Background colors	❌ No	Use white/transparent
3D effects	❌ No (distorts data)	Remove immediately
Decorative icons	❌ No	Remove
Drop shadows	❌ No	Remove

What does a high vs. low data-ink ratio look like?

The same bar or line chart can range from roughly 0.3 (mostly decoration) to 0.8 (mostly data) depending on how many non-data elements survive the cleanup.

Example 1: Bar Chart

❌ Low Data-Ink Ratio (Cluttered)

Problems:
- Heavy black border around entire chart
- Dark gray background
- Thick grid lines every 10 units
- 3D bars with shadows
- Decorative icons on bars
- Redundant legend (colors match labels)
- Unnecessary decimal places (253.00 vs 253)

Data-Ink Ratio: ~0.3 (only 30% shows data)

✅ High Data-Ink Ratio (Clean)

Improvements:
- No border (axes define the space)
- White background
- Light gray grid lines (or removed entirely)
- Flat 2D bars
- Direct labels on bars (no legend needed)
- Rounded numbers (253, not 253.00)

Data-Ink Ratio: ~0.8 (80% shows data)
Result: Easier to read, focuses attention on data

Example 2: Line Chart

❌ Low Data-Ink Ratio

Problems:
- Both major and minor grid lines
- Heavy axis lines
- Markers on every data point (100+ points)
- Thick lines (3px+)
- Legend box with border
- Both left and right Y-axes (showing same scale)

Focus: Split between data and decoration

✅ High Data-Ink Ratio

Improvements:
- Minimal grid lines (or none, let axis labels suffice)
- Thin axis lines or remove entirely
- No markers (or only on key points)
- 2px lines
- Direct labels on lines (no legend)
- Single Y-axis

Focus: Entirely on the trend

How do I raise a chart's data-ink ratio step by step?

Work through five passes in order: delete the obvious junk, simplify colors, reduce redundant elements, lighten what's left, then emphasize the data itself.

Step 1: Remove the Obvious Junk

Delete immediately:
❌ 3D effects and shadows
❌ Decorative borders
❌ Background patterns/images
❌ Unnecessary animation
❌ Chartjunk (pictographs, excessive decoration)

These add zero information.

Step 2: Simplify Grid Lines

Grid line strategy:

Option 1: Remove entirely
- Best for: Simple charts, few data points
- Let axis labels provide reference

Option 2: Lighten drastically
- Color: Light gray (#E0E0E0 or similar)
- Weight: 1px or less
- Frequency: Every 2-3 axis labels, not every one

Option 3: Horizontal only
- Keep horizontal lines (easier to read values)
- Remove vertical lines (time periods are obvious)

Step 3: Reduce Redundancy

Common redundancies:

❌ Legend + colored bars + data labels
→ Choose 2 of 3 (usually bars + labels)

❌ Y-axis + data labels on bars
→ Choose one (labels if values matter, axis if pattern matters)

❌ Both axes showing same scale
→ Remove right axis

❌ Every data point labeled
→ Label key points only (min, max, start, end)

Step 4: Lighten Non-Data Elements

Make supporting elements subtle:

Axis lines:
- Thin (1px)
- Light gray
- Or remove entirely (let labels define space)

Grid lines:
- Very light gray (#F0F0F0)
- Dashed or dotted (less prominent)

Text:
- Dark gray instead of black (#333 vs #000)
- Smaller font for secondary labels

Step 5: Emphasize the Data

Now that clutter is gone, make data stand out:

✅ Bolder colors for data elements
✅ Slightly thicker lines (but not too thick)
✅ Highlight key data points
✅ Use color strategically (one accent color)

The data should be the darkest, most prominent element

When should I keep grid lines, borders, or background color?

High data-ink ratio doesn't mean stripping everything. Context matters:

Keep Grid Lines When:

• ✅ Precise value reading is critical
• ✅ Audience needs to reference specific numbers
• ✅ Comparing many series simultaneously

Keep Borders When:

• ✅ Chart is on busy dashboard (needs definition)
• ✅ Multiple charts side-by-side need separation

Keep Background Color When:

• ✅ Creating visual hierarchy (card-based dashboards)
• ✅ Distinguishing chart from surrounding content

Guideline: If removing an element makes the chart harder to understand, keep it. The goal is clarity, not minimalism for its own sake.

Common Mistakes

❌ Mistake #1: Over-Simplification

Too far: Removing all axis labels and grid lines
Result: No way to read actual values

Fix: Keep essential reference points (axis labels minimum)

❌ Mistake #2: Inconsistent Application

Bad: Clean charts in one dashboard, cluttered in another
Result: Inconsistent user experience

Fix: Establish and follow organization-wide standards

❌ Mistake #3: Removing Color When It's Data

Bad: Making all bars gray to "increase data-ink ratio"
Result: Can't distinguish categories

Fix: Color IS data when it represents categories/values. Keep it.

How do I raise the data-ink ratio in Excel or Tableau/Power BI?

Both tools default to heavier styling than you need — the fix is mostly turning off fills, borders, and gridlines in the formatting panel.

Excel/Google Sheets Charts

Quick improvements:

1. Format Chart Area
   - Fill: None
   - Border: None

2. Format Plot Area
   - Fill: None
   - Border: None

3. Format Gridlines
   - Color: Light gray
   - Width: 0.5pt
   - Or delete entirely

4. Format Axes
   - Line color: Light gray
   - Width: 1pt

5. Remove Legend
   - Add data labels directly instead

Result: Instant 50% improvement in data-ink ratio

Tableau/Power BI

Default settings add lots of non-data ink. Customize:

1. Format > Borders
   - Row/Column dividers: None
   - Cell borders: None

2. Format > Gridlines
   - Make very light or remove

3. Format > Shading
   - Remove row/column shading

4. Use direct labels instead of legends

What five changes give the biggest data-ink improvement?

These 5 changes give you 80% of the benefit:

1. Remove chart borders (or make them very light)
2. Lighten grid lines to barely visible (#F0F0F0) or remove
3. Remove 3D effects everywhere
4. Use white backgrounds instead of gray/colored
5. Direct label instead of legends when possible

How do I measure my chart's actual data-ink ratio?

Two quick checks work without any special tools: the screenshot test (highlight data pixels vs. total pixels) and the squint test (what stands out when the chart is blurry).

The Screenshot Test

1. Take a screenshot of your chart
2. Open in image editor
3. Highlight all pixels that represent actual data
4. What percentage is highlighted?

Targets:

• 60%+ = Good
• 70%+ = Very good
• 80%+ = Excellent

The Squint Test

Squint at your chart. What stands out?

• ✅ Good: The data elements (bars, lines, points)
• ❌ Bad: Grid lines, borders, backgrounds

What's a quick checklist for data-ink ratio?

Run any finished chart through this list before shipping it:

• ☐ Chart border removed or very light
• ☐ Grid lines lightened (#E0E0E0+) or removed
• ☐ No 3D effects or drop shadows
• ☐ White or transparent background
• ☐ No decorative elements (icons, patterns, textures)
• ☐ Legend removed in favor of direct labels (if possible)
• ☐ Axis lines thin (1px) or removed
• ☐ No redundant data representation
• ☐ Data elements are most prominent visual element
• ☐ Every remaining element serves a purpose

← Back to Visualization Tips