A Cheat Sheet for Choosing the Right Data Visualizations
Discover the best practices for designing these 10 popular charts.
Today, organizations of all industries and sizes use data to tap into the wellness of their business. Anyone who can easily see trends and spot problems can make smarter decisions. Therefore, you have to visualize your data if you want to be competitive.
Data visualizations are images that show measurement and activity. They’re key components of business presentations and they’re the building blocks of dashboards. But despite their ubiquity, not everyone knows how to select and format them properly. And if you can’t visualize your data well, your colleagues can’t uncover any insights that lead to growth.
Every successful data visualization begins with a clear understanding of what you want to analyze. Once you realize the true purpose of these data visualizations — and avoid some common pitfalls with using them — you can effectively illustrate any type of information you want.
1: SINGLE VALUE
When you’re presenting high-level updates to your readers, you may only need to share single values. Totals, averages, and rates are convenient metrics because they can summarize entire datasets in one number. You can give them more context with visual indicators, such as their percent difference from a previous period. Consider rounding and abbreviating these figures to make them more readable.
2: BAR CHART
As information designer Bill Shander puts it, “You should almost always ask yourself, ‘Why should I not do a bar chart?’” A bar chart breaks a measure down into separate categories, which is a basic objective of data analysis. A stacked bar chart breaks everything down further into sub-categories. Regardless of which variation you choose, you can orient the bars either vertically or horizontally. In all cases, however, you must start the y-axis at zero. Otherwise, you will exaggerate the differences between your bars and distort their relative size.
3: PIE CHART
Since a pie chart consists of parts of a whole, its values have to add up to 100 percent. Yet people often misuse this data visualization to show several unrelated measures in one place. They also create pie charts with numerous slices, which is bad practice considering that humans can’t accurately interpret 2D angles. As a result, you should only use a pie chart if you’re splitting a measure into five parts or less. Anything more than that probably belongs in a bar chart.
4: LINE CHART
A line chart connects points together to demonstrate how their values change over time (e.g. years, months, or days). It’s ideal for trend analysis because the line’s overall shape can indicate increases, decreases, fluctuations, and other patterns. A line chart comes in handy for assessing relationships between different measures, like the real GDP versus the real median household income. In contrast to a bar chart, it’s not mandatory to start the y-axis at zero.
5: SPARKLINE
Much like a line chart, a sparkline illustrates change over time. Its shape implies how stable, cyclical, or volatile a trend has been. But unlike a line chart, it lacks a labeled axis and plot points are reserved for the most recent or high and low values. A sparkline offers a minimal representation of past activity and it can quickly lend historical context to your results.
6: BULLET CHART
A bullet chart is essentially a bar chart, except it uses one bar and it has graded levels. It tracks a single measure against a target value and performance ranges (e.g. bad, satisfactory, good). A bullet chart is a simple alternative to a gauge, which is a data visualization that’s borrowed from a literal car dashboard.
7: SCATTER PLOT
A scatter plot’s sole purpose is to demonstrate whether or not a correlation exists between two variables — one plotted along the x-axis and another plotted along the y-axis. When they increase together, the correlation is positive. When one increases and the other one decreases, the correlation is negative. If there’s no discernible pattern either way, there’s no correlation. If you can detect a positive or negative correlation in your scatter plot, consider inserting a trend line to highlight this relationship.
8: BOX PLOT
A box plot exposes the distribution of a set of numbers. The first, second, and third quartile form the shape of the box, while the outside lines extend to the maximum and minimum values. Any high or low outliers appear as dots outside their respective ends. A box plot is a practical data visualization for test scores, salaries, load times, and anything else you can’t adequately capture in a single value (like an average or total).
9: CHOROPLETH & SYMBOL MAPS
Derived from the Greek words for “region” and “multitude,” a choropleth map uses color to indicate value in certain places. It’s ideal for showing variance within a geographic territory. The data in a choropleth map has to be relative to a specific area, so it’s important to stick with regional metrics like population density. A lot of people mistakenly encode this data visualization with whole numbers like population size — in which case, they need to use a symbol map.
10: TABLE
A table arranges data into columns and rows. The format makes it easier for readers to look up information based on a specific combination of factors. It’s especially useful for financial analysts who depend on detailed reports with full numbers. A table is a nice addition to data visualizations that only provide a “snapshot” of information. If you create a table with multiple columns or wide rows, introduce gridlines or alternating colors to help your readers scan the data.
SOME RESOURCES & TIPS
You have an array of options when it comes to visualizing your data. Microsoft Excel and Google Sheets have built-in data visualization features, but there are also enterprise BI tools like Tableau, Microsoft Power BI, Looker, and Domo. D3.js is an open-source JavaScript library that’s a good fit for anyone who knows how to code. Meanwhile Plot.ly, RAWGraphs, and datavisu.al are some beautiful platforms that are user-friendly and free to try.
Be prepared to apply some finishing touches to make your data visualizations truly effective. Don’t forget to title your charts, label their axes, and include legends that explain your symbols or colors. Avoid busy color schemes and be judicious with red since too much of it can needlessly grab your readers’ attention. But no matter what you do…never, ever visualize your data in 3D! This effect adds extra colors, shapes, and angles that will only make your data visualization harder to decipher.
When you design a data visualization, it’s your job to ensure that it’s meaningful and easy to interpret. A poor data visualization can confuse or mislead your audience, while a good one can spur them to action and guide them to success.