Visualizations That Communicate

Visualization Purpose

In the previous lesson, we explored rapid data exploration. Now let’s build on that foundation. Charts exist to communicate, not to impress.

A good visualization makes a pattern obvious. A bad visualization makes your audience work to understand what they’re seeing.

The question is never “what chart can I make?” It’s “what do I need to show, and what’s the clearest way to show it?”

Matching Charts to Messages

Different chart types show different relationships:

Comparison: Bar Charts

Use when: Comparing values across categories.

Good for:

• Sales by region
• Performance by team member
• Market share by competitor

Rule: Horizontal bars for many categories or long labels. Vertical for few categories.

Trend Over Time: Line Charts

Use when: Showing how something changes over time.

Good for:

• Revenue by month
• User growth over quarters
• Metric performance over time

Rule: Time goes on the x-axis. Limit to 3-4 lines maximum.

Part-to-Whole: Pie/Donut Charts

Use when: Showing proportions that add to 100%.

Good for:

• Market share breakdown
• Budget allocation
• Customer segment distribution

Rule: Maximum 5-6 slices. Order by size. Use sparingly—bar charts often work better.

Distribution: Histogram

Use when: Showing how values are distributed.

Good for:

• Order value distribution
• Age distribution of customers
• Response time distribution

Relationship: Scatter Plot

Use when: Showing correlation between two variables.

Good for:

• Price vs. quantity sold
• Ad spend vs. revenue
• Size vs. performance

The Chart Selection Guide

AI-Assisted Visualization

Use AI to suggest and describe visualizations:

I have data showing:
- Monthly revenue for the past 12 months
- Broken down by 3 product categories
- For 4 different regions

I need to show: Revenue trends by product category

Suggest the best chart type and describe:
1. What chart type and why
2. What goes on each axis
3. How to handle the multiple dimensions (product, region)
4. Any design recommendations

If your tool supports it, have AI generate the actual chart or the code to create it.

Visualization Best Practices

1. One Chart, One Message

Bad: A chart trying to show sales trend AND regional comparison AND growth rate.

Good: Separate charts, each with a clear point.

2. Title as Insight

Bad title: “Sales Data” Good title: “Sales Grew 23% in Q3, Driven by Enterprise Segment”

Your title should state what the viewer should take away.

Quick check: Before moving on, can you recall the key concept we just covered? Try to explain it in your own words before continuing.

3. Start Y-Axis at Zero (Usually)

Truncated axes exaggerate differences. Unless there’s a good reason, start at zero.

4. Remove Clutter

Delete anything that doesn’t add information:

• Gridlines (reduce or remove)
• 3D effects (always remove)
• Excessive legend when labels work
• Unnecessary decimals

5. Use Color Intentionally

• Highlight: Use color to draw attention to key data
• Categorize: Consistent colors for categories across charts
• Don’t overdo: 3-5 colors maximum

6. Provide Context

Add reference lines for:

• Targets or goals
• Historical averages
• Industry benchmarks

Common Visualization Mistakes

Mistake: Pie Charts with Too Many Slices

If you have more than 5-6 categories, use a bar chart instead.

Mistake: Dual Y-Axes

Two y-axes confuse viewers. If you must compare different scales, use separate charts.

Mistake: Truncated Axes That Mislead

A y-axis starting at 95 instead of 0 makes a 3% change look dramatic.

Mistake: Too Much Data

20 lines on one line chart is unreadable. Aggregate, filter, or use small multiples.

Mistake: Wrong Chart Type

A pie chart for trend data. A line chart for unordered categories. Match chart to message.

Creating Charts with AI

Describing Charts for AI Generation

Create a bar chart showing:
- Data: Revenue by region (North: $1.2M, South: $0.8M, East: $1.5M, West: $0.9M)
- Title: "East Region Leads Revenue in Q3"
- Highlight: East region bar in blue, others in gray
- Include: Target line at $1.0M
- Format: Clean, minimal gridlines

Getting Chart Code from AI

Write Python code (matplotlib) to create a line chart showing:
- X-axis: Months (Jan-Dec)
- Y-axis: Revenue
- Lines: 3 product categories
- Title: "[Descriptive title]"
- Style: Clean, professional

Data:
[Paste your data]

Exercise: Choose the Right Chart

For each scenario, what chart type would you use?

1. Comparing this quarter’s sales across 5 sales reps
2. Showing how website traffic has changed over 24 months
3. Breaking down where marketing budget is spent (5 channels)
4. Showing the relationship between customer age and order value
5. Comparing this year’s monthly revenue to last year’s

Key Takeaways

• Charts exist to communicate, not impress—choose based on the message
• Match chart type to relationship: comparison→bar, trend→line, proportion→pie
• One chart, one message; title should state the insight
• Remove clutter: gridlines, 3D effects, excessive colors
• Use AI to suggest visualization approaches and generate chart code
• Common mistakes: too many categories, dual axes, truncated scales, wrong chart type

Next: extracting meaningful insights from your analysis.

Up next: In the next lesson, we’ll dive into Finding Insights, Not Just Numbers.