Choosing the Right Chart Type for Every Message

Chart types should not be chosen by taste or appearance. Each form answers one type of question well and others poorly. A trend in a pie chart, a comparison with an unreadable axis, or a profile in a numeric table can all contain correct data while leaving the message distorted or invisible.

The mapping from question to form is manageable. Seven presentation forms cover the foundations of most dashboards.

---

This article covers the following presentation forms:

• KPI element
• line chart
• column and bar chart
• pie and donut chart
• polar and radar chart
• table
• text

The overview below explains what each element shows, when it works well, and when it does not. It also covers common mistakes, corrections, and the rules that turn individual charts into a coherent dashboard.

---

The question determines the form

Before choosing a chart type, be clear about the question it should answer. Five basic questions cover almost everything: How high is a value right now? How is it developing over time? What is larger or smaller? What makes up the whole? And how balanced is something across several dimensions?

Perception is the decisive factor. Each form encodes values differently—as position, length, angle, or area—and people read these encodings with different levels of precision.

Fact

Research into graphical perception shows that people read comparisons based on position and length, such as in column and line charts, much more accurately than comparisons based on angles and areas, as required by pie charts.

Source

This leads to the central rule: whenever precision matters, choose the form that translates your message into position or length. Use angle and area only when a broad impression is enough. The table summarizes the mapping; the following sections explain each form.

Single values: the KPI element

A KPI element displays one number prominently, such as monthly revenue, open tickets, or conversion rate. It answers the most common dashboard question: Where are we right now? Alternatively: What is the average?

Context makes the number useful. A previous-period value, target, or trend direction creates a quick overview. The strengths are clearest when the element is:

• paired with a comparison or target
• labeled with a unit and period
• used sparingly

Five to seven KPI elements per view are a useful guideline. More are rarely important. Deciding which metrics deserve this space is a definition question before it is a design question. See Data and KPIs in BI.

Do not use a KPI element when the development itself is the message. A trend arrow does not replace a time series: it shows direction, but not scale, pace, or turning points.

Trends: the line chart

A line chart is well suited to anything that develops over time: revenue by month, visitors by week, or costs by day. Time runs along the horizontal axis, and the eye reads slopes, turning points, and seasonal patterns with little effort—the exact signals that disappear in detailed tables.

Four rules keep trends readable:

• Use few lines: Three to five metrics per chart allow the lines and axes to be read comfortably.
• Use the same units: Lines within one chart should share a unit. Additional units create ambiguity.
• Choose meaningful granularity: Daily values provide detail; monthly values provide an overview. Match the resolution to the question.
• Label exceptional events: A campaign launch, price increase, or public holiday gives viewers essential context.

Do not use a line chart for categories without a time relationship. A line across “Campaign A, B, C” suggests a development that does not exist. Use columns or bars instead.

Comparisons: column and bar charts

Columns and bars are highly effective for comparisons outside time: revenue by product line, leads by channel, or effort by project. They translate values into lengths. Three variants have distinct purposes.

Columns for a few categories

Vertical columns work for a handful of categories with short labels, such as four regions or six product groups. Sort by value rather than alphabetically when the order has no meaning of its own. The ranking is usually an important part of the message.

Bars for many categories and long names

Horizontal bars are preferable once there are more than six to eight categories or the labels are long. Campaign names, page titles, and customer names remain readable horizontally. Sort by value, limit the view to the top entries where appropriate, and group the remainder.

Grouped and stacked

Grouped columns place two or three series side by side within each category, such as this year versus last year by quarter. Stacked columns also show the total, but have a known weakness: only the bottom segment starts from a common baseline and can be compared precisely. Middle segments “float” and permit only rough comparison. Use stacked columns when the total adds valuable context. Use grouped columns when the individual values are the focus.

Shares: pie and donut charts

Pie and donut charts answer exactly one question: How much does each part contribute to the whole? Examples include revenue share by region, budget allocation by channel, or the share of returning customers. They work where the segments add up to one hundred percent.

The eye reads angles imprecisely, which leads to three rules:

1. Use no more than five or six segments, and group smaller positions as “Other.”
2. If segments are similarly sized and their ranking matters, use bars; the difference will be easier to read.
3. A pie chart for values that do not add up to a whole is formally incorrect.

Do not use pie charts for developments over time or precise comparisons across many categories.

Profiles: polar and radar charts

Radar and polar charts show several dimensions of one entity: a campaign across five assessment criteria, a location across six performance dimensions, or a vendor against a criteria catalog. They are strong when the question is “How balanced is this?” because patterns and outliers in the profile stand out.

Three rules apply:

1. Use comparable scales on every axis, or the shape will misrepresent the relationship between values.
2. Overlay no more than two, at most three, profiles; beyond that, the display becomes cluttered.
3. Compare the shape rather than the area. The enclosed area grows disproportionately and makes small differences look large.

Do not use radar charts when precise comparisons are needed for each dimension. Radar is a form for the overall impression, not decimal-level accuracy.

Details: the table element

Tables are indispensable for detailed insight in dashboards. They combine metrics and dimensions in a grid and provide more information than any chart: What are the exact values, and what lies behind them?

Their purpose is depth—as a detail layer beneath charts, for root-cause questions such as “which campaigns exactly?”, and for anything that users need to look up rather than scan. Four rules apply:

1. Right-align numbers and format them consistently.
2. Use a few deliberately selected columns instead of every available field.
3. Set a default sort that answers the most important question.
4. Use highlights sparingly and only for what truly matters.

Do not use a table as the first view for trend or comparison questions. A table complements charts; it does not replace them.

Context: the text element

Text elements show no data, yet belong in every good dashboard. They communicate how to read the numbers, such as “figures as of the previous day,” define disputed metrics, note exceptional effects, and record the decision resulting from the numbers.

This context becomes especially important when sharing with recipients outside the specialist department. One sentence beside a chart can prevent ten follow-up questions. The boundary is equally clear: a dashboard is not a report. Short, stable guidance belongs in the dashboard; changing prose and analysis belong in the accompanying format. Reporting, dashboards, and analysis explains when each format is appropriate.

Common mistakes—and the quick correction

This table summarizes frequent visualization mistakes and the advantage of the corresponding correction:

These mistakes usually arise because the form was chosen by appearance rather than by question. The correction is often simply switching to the appropriate form.

From an individual chart to a coherent dashboard

The interaction between elements matters at least as much as the choice of each individual element. Three principles make dashboards easier to read:

1. Consistency: Use forms consistently. Repetition makes a dashboard easier to read; variety has no value by itself. Apply the same principle to colors, labels, grids, and axes. Use variation selectively where it creates value, particularly for emphasis.
2. Hierarchy: Use a consistent structure within and across dashboards. For example, place summary KPI elements at the top, trends and comparisons in the middle, and details in a table below. The viewer's eye then moves naturally from “Where are we?” to “Why?”
3. Reduction: Every element should answer a question that users actually ask. Too much information overloads the dashboard. Each element needs a clear purpose tied to the dashboard's goal and the decisions derived from it.

Build effective dashboards explores the overall structure in more depth.

Frequently asked questions about chart types

Which chart is suitable for comparing months across a year?

A line chart makes developments across the year—trends, seasonal patterns, and turning points—immediately visible. For a direct comparison between individual months, such as December this year versus December last year, grouped columns are the better choice. A pie chart is not suitable for time comparisons.

How many charts belong on a dashboard?

The number depends on the use case and should not be fixed rigidly. Five to nine elements per view can serve as a rough guideline for focused dashboards. That is usually enough to answer one guiding question. If more are needed, check whether two separate questions have been mixed together. A second view may serve the goal better than a fuller grid.

What role do scatter plots and heat maps play?

They are specialists for exploratory and more complex analyses. Scatter plots examine relationships between two metrics; heat maps reveal density and patterns. They are generally used for targeted analysis rather than dashboards that answer recurring questions.

Conclusion

Choosing chart types correctly means determining the question first and the form second: single value, trend, comparison, share, profile, detail, or context. Each task has forms that support it well and forms that do not.

Analyze the questions your team asks after viewing a dashboard. They reveal where ambiguity occurs and help you prevent it deliberately.