Why Bubble Charts Require Extra Caution

1. We are bad at comparing area

A bubble’s size represents value using area, not radius.
But our eyes don’t intuitively judge area correctly.

A bubble that’s 2× the value doesn’t look 2× bigger — it often looks much less.

This makes bubble charts imprecise when:

• Values need strict comparison
• Differences are subtle
• The audience needs quick interpretation

---

2. Slight size differences look identical

Two bubbles with values like 45 and 50 can appear almost the same.
Important differences get visually lost.

---

3. Sorting or ordering bubbles is nearly impossible

You can’t visually “order” bubbles at a glance the way you can with bars.

---

4. Overlapping bubbles hide information

If values are close or densely plotted:

• Bubbles overlap
• Labels collide
• Smaller bubbles disappear
• Interactions become confusing

Scatter plots with size encoding are especially sensitive to this.

---

5. Bubble scales are often misleading (if not calculated carefully)

Many tools mistakenly scale bubble diameter instead of area, which doubles the visual distortion.

Correct scaling requires:
area ∝ value, not diameter.

If a charting tool or library doesn’t get this right, the chart becomes misleading.

---

So when should bubbles actually be used?

Use bubbles only when size is not the primary comparison tool, but an extra dimension that enriches a relationship.

Good use case:

✔ Showing three variables at once
(e.g., scatter plot: x = revenue, y = profit, bubble size = number of customers)

✔ Showing “big picture” patterns where exact precision is not required
(e.g., clustering)

✔ When differences in magnitude are large
(e.g., 10 vs 500, where size distinction is obvious)

---

When to avoid bubbles

❌ When you need exact comparisons
❌ When values are very close
❌ When the audience is not analytical
❌ When you have more than ~30 bubbles
❌ When the chart looks crowded
❌ When size is the main message (bars are better)

---

Cleaner alternatives to bubble charts

• Scatter plot (without bubble sizing)
• Bar or column chart
• Dot plot
• Treemap (if you truly need area)
• Heatmap (if color is a better encoding)

---

In short:

Bubble charts are powerful but prone to misinterpretation.
That’s why I said “with care” in A Practical Formula for Choosing the Right Data Visualization & Visual Encoding Channels — they’re a tool worth using intentionally, not casually.