Chart Types and When to Use Them

The Wrong Chart Problem

A marketing manager presents quarterly results to the executive team. She uses a pie chart to show revenue by region: North America 42%, Europe 31%, Asia 18%, Other 9%.

The CEO asks: “How has the regional mix changed over the last four quarters?”

She shows four pie charts side by side. Four circles, sixteen slices. Can you tell if Europe grew from 28% to 31%? Can you see whether Asia’s growth is accelerating?

No. You can’t. Because pie charts are terrible at showing change over time.

The right chart for that question is a stacked bar chart (one bar per quarter, colored by region) or a line chart (one line per region). Either would answer the CEO’s question at a glance.

Choosing the right chart is the single most impactful decision in data visualization. Let’s make sure you always get it right.

The Chart Decision Framework

Every chart selection comes down to answering one question: What relationship am I showing?

Let’s explore the most common types in detail.

Bar Charts: The Workhorse

Bar charts are the most versatile chart type. When in doubt, a bar chart is rarely wrong.

Use for: Comparing values across categories

Vertical bars when categories are on the x-axis (time periods, product names):

Revenue by Quarter:
Q1: ████████░░ $245K
Q2: █████████░ $267K
Q3: ██████████ $312K
Q4: ███████████ $358K

Horizontal bars when category names are long or you’re showing a ranking:

Top Customer Complaints:
Slow shipping    ██████████████ 342
Wrong item       ████████████   287
Damaged package  ████████       201
Late delivery    ██████         156
Missing items    ████           98

Common mistakes:

• Starting the y-axis at a value other than zero (distorts comparison)
• Using too many colors when one color with varying intensity works
• Not sorting bars in meaningful order (alphabetical is rarely the most useful)

Line Charts: Showing Trends

Line charts are for continuous data, usually time series.

Use for: Showing trends, rates of change, and comparisons over time

When lines work:

• Monthly revenue for the past 2 years (shows trend)
• Daily website visitors (shows patterns and anomalies)
• Multiple product lines over time (shows relative performance)

When lines don’t work:

• Comparing 3 categories at a single point in time (use bars)
• Showing parts of a whole (use stacked bars)
• Data with only 2-3 time points (bars are clearer)

Pro tip: When you have multiple lines, don’t use a legend that forces the viewer to bounce between chart and legend. Label the lines directly on the chart, at their end points.

Scatter Plots: Finding Relationships

Scatter plots are underused and incredibly powerful.

Use for: Exploring relationships between two variables

Each dot = one customer
X-axis = Amount spent on marketing
Y-axis = Revenue generated

Pattern: dots trending upward = more marketing → more revenue
Outlier: one dot high on Y but low on X = organic success
Cluster: group of dots at low X, low Y = needs investigation

When to use scatter plots:

• Does ad spend correlate with revenue?
• Do experienced employees produce more output?
• Is there a relationship between page load time and bounce rate?

Enhancement: Add a trend line to make the relationship explicit. AI can calculate and suggest this.

The Pie Chart Debate

Pie charts are controversial. Here’s the rule:

Use pie charts ONLY when:

• You have 2-4 categories
• You want to emphasize that parts make up 100%
• The differences between slices are obvious (one slice dominates)

Never use pie charts when:

• You have more than 5 categories
• Slices are similar in size (can you tell 23% from 26% by angle?)
• You need to compare across time periods
• You have negative values

Alternative: A horizontal bar chart showing percentages communicates the same information more accurately. Your viewer can compare bar lengths far more precisely than slice angles.

Quick Check: Pick the Chart

For each scenario, choose the best chart type:

1. Showing your company’s market share (35%) vs. three competitors Answer: Pie chart (4 slices, part-to-whole, one dominant slice) or horizontal bar chart

2. Tracking customer satisfaction scores monthly for 18 months Answer: Line chart (continuous time series, shows trend)

3. Comparing feature usage across five user segments Answer: Grouped bar chart (comparing categories across groups)

4. Showing how website visitors flow from homepage to purchase Answer: Funnel chart or Sankey diagram (flow/conversion)

Using AI for Chart Selection

Here’s the prompt that saves you from chart selection uncertainty:

I need to visualize this data. Help me choose the right
chart type.

Data description:
- 12 months of sales data
- Broken down by 4 product categories
- I want to show that Category B grew 300% while others
  were flat
- Audience: Executive team in a quarterly review
- Will be shown on a presentation slide (not interactive)

What chart type should I use? Explain your reasoning.
If there are multiple good options, compare them.

AI might respond: “A multi-line chart is your best option. It shows all four category trends simultaneously, and the steep upward slope of Category B will visually dominate against the flat lines of the others—making your ‘300% growth’ message immediately obvious. A grouped bar chart would also work but requires 48 bars (12 months x 4 categories), which is too dense for a presentation slide.”

This kind of reasoned recommendation is exactly what you need.

Beyond Basic Charts

For specialized situations:

Heatmaps — When you have too many data points for individual marks. Great for showing patterns in website analytics (activity by day of week and hour of day).

Treemaps — For hierarchical data with many categories. Better than pie charts for budgets with 20+ line items.

Waterfall charts — For showing how a starting value is affected by positive and negative changes. Perfect for “how did we get from $1M budget to $750K remaining?”

Small multiples — Instead of cramming everything into one chart, show a grid of small, identical charts, each showing one subset. Ten small line charts are clearer than one chart with ten lines.

Sparklines — Tiny inline charts embedded in tables. They add trend context to tabular data without requiring a separate chart.

Chart Selection Cheat Sheet

Ask yourself these three questions:

1. How many variables? One → bar/pie. Two → scatter/line. Three+ → consider small multiples
2. Is time involved? Yes → line or area chart. No → bar, scatter, or pie
3. What’s the message? Comparison → bars. Trend → lines. Composition → stacked/pie. Relationship → scatter

When still unsure, describe your data and message to AI. It’ll recommend a chart type faster than you can look up best practices.

Practical Exercise

Take a dataset you work with and create three different visualizations of the same data:

1. The “obvious” chart (whatever your tool defaults to)
2. The chart AI recommends for your specific message
3. An alternative chart type for a different angle on the same data

Compare them. Which communicates the message most clearly? Which would make sense to someone seeing it for the first time?

Key Takeaways

• Chart selection is the highest-impact decision in data visualization
• Match the chart type to the relationship you’re showing: comparison, trend, composition, distribution, or correlation
• Pie charts are only appropriate for 2-4 clearly different categories
• Line charts are for time series; bar charts are for categorical comparisons
• Use AI to get reasoned chart type recommendations by describing your data, message, and audience
• When in doubt, a bar chart is rarely wrong

Next up: design principles. You’ve chosen the right chart type—now let’s make it clear, clean, and professional.

Up next: In the next lesson, we’ll dive into Designing Clear Visualizations with AI.