5 Steps to Improve Data Visualizations

Subscribe at the end of the post to receive

the eBook for free


As Scott Berinato suggested,[1] making good charts requires you to adopt “visual thinking.” Adopting visual thinking means understanding how we see and interpret visual elements. First, we don’t follow a specific order like in reading a text. Of course, we tend to start by looking at the upper-left corner, but our eyes go where visual elements stand out (bold text, highlighted elements, peaks and valleys in graphs, intersections, outliers, striking colors). Second, we can only see a limited number of visual elements at once. As a rule of thumb, starting with five to ten elements, we are no longer able to clearly identify individual elements, but we start to see global or average patterns. Third, we always try to seek meaning, identify patterns, and make conclusions of what we see. Therefore, you have to carefully use colors, names, and positions because you may unintentionally create misleading associations, trends, or patterns. For example, when you show a trend in time, the choice of the time frame can completely change the message you show.

1) Choose the best available solution

The first phase when building a chart is thinking about what you want to communicate. Of course, you will already have analyzed your data, but you should try to forget about it for a moment. Don’t be limited by the constraints of your data. Also think about where, when, and to whom the chart will be presented. Talk to someone, the conversation can add additional information and can improve the message you want to communicate. It is important to clearly understand the reasons why you are going to create and present this chart.

In the second phase, you start sketching it, preferably on a paper. At this point, you are visualizing your chart and this means that you have thought about a specific kind of chart. To help you in this task, Scott Berinato suggests writing down what you want to show and taking notice of the keywords used. In his book Good Charts, he presents an interesting example: “I want to COMPARE THE NUMBER of job posting to hires to SEE what THE RATIO is for DIFFERENT TYPES of jobs.” The keywords highlighted in this phrase suggest you use a chart to compare ratios of different categories.

Example of chart comparing job hires and job postings (Source: Adapted from Scott Berinato, Good Charts)

The author also suggests a very useful template you can use as guideline for choosing the most appropriate chart based on the keywords you identify in your phrase.

How to choose a type of chart based on keywords (Source: Adapted from Scott Berinato, Good Charts)

2) Use a consistent structure and visual order

If your chart is well structured and elements are well positioned, you improve both its readability and aesthetic. The advantages of an easier-to-understand chart are quite clear, but what about aesthetics? A “beautiful” graph has several advantages; for example, it grabs the attention of the audience, it is more effortless to read, it seems more professional, and conclusions are more convincing.

Start by applying a standard structure with a specific order and weight: for example, title 12%, subtitle 8%, field 75%, and source line 5%. Finally, to improve the visual order of your chart it is advisable to respect alignment and white spaces as in the example below.

Example about how to visually improve a chart (Source: Adapted from Scott Berinato, Good Charts)

3) Lead your audience toward the main idea

The key elements you can work with to lead your audience toward the main idea are contrast, positioning, and additional elements. Contrast works well when we have a few types of elements of contrast to compare (ideally two colors, two shapes, etc.). It is easier to spot a blue element among many gray ones, but it is more difficult to spot it among many different colors.

Besides contrast, positioning has been used in the following example since the attributes are presented in descending order starting from our company’s performance. The audience will start looking at the graph by the upper-left element. In this example, the items are ordered vertically; in other charts you may choose to order them horizontally. Finally, additional elements like text can help convey the main idea, for example, by making it explicit in the title (e.g., “Sales are decreasing in the last quarter”) or adding additional information.

Example of positioning

Another positioning technique is the modification of reference points to emphasize the main idea. For example, you can remove reference points by eliminating some less important details to make your point clearer. In the following example, age groups have been reduced and attributes have been grouped by age. You will notice that the second chart clearly shows the divide between young and elder people.

Example of removing reference points

4) Remove unnecessary elements

Each element must fulfill two requirements to be kept in the visual. First, it must be useful to communicate your idea, or, put another way, without this element you can’t communicate your idea, or it will be less clear. Second, if other elements play the same role, it must be the most effective. If this element is necessary and the most effective, ask yourself if you can further improve it by making it more effective or simplifying it.

How to choose which elements to keep or eliminate in a chart (Source: Adapted from Scott Berinato, Good Charts)

Some practical recommendations are:

  • Remove borders, gridlines, and data markers, or at least make them less visible.
  • Clean up axis labels: The text must be horizontal, names must have an appropriate length (you may use abbreviations), and numbers must have the appropriate level of detail (don’t use decimals if the scale is about units, tens, hundreds, etc.).
  • Keep axes demarcations at minimum or remove the whole axis if not necessary (it would be redundant if you show data labels).
  • Legend: You can label series or categories directly in the chart field instead of creating a legend below it.
  • Don’t add 3D to graphs.
  • Don’t repeat axes labels in titles or subtitles; instead, exploit them for showing the question or the answer associated with your visual.
  • Push to the background all the secondary elements.

5) Make it easier for your audience

In the following example, the graph has been simplified and improved as you can see in the second one. First, having a legend with five colors implies an extra effort for your audience, which you can reduce by using a common x-axis with the different years. Second, you can pass from twenty-five bars to four lines where it is easier to highlight what the graph was meant to highlight: the number of marriages for those who have a bachelor’s degree or more. Finally, the decimal points have been removed. They didn’t convey any useful additional information.

How to simplify a chart by reducing the number of colors, elements, and decimals (Source: Adapted from Cole Nussbaumer Knaflic, Storytelling With Data)

Besides simplification, try to be consistent. If you start your presentation with some color pattern, stick to it, otherwise, the audience may be misled. In addition, if you identify a category with a specific color, maintain it throughout the presentation (e.g., the color used to identify a specific competitor). Remember that due to the limits of our short-term memory, we can only keep in mind three to nine visuals; so the more colors (or other elements) you use, the more effort is needed. The only elements that can sometimes exceed this limit are bars and lines, which we can easily compare. Another limit is the number of tonalities of a color we can distinguish.

Order and color have been modified to respect conventions (Source: Adapted from Scott Berinato, Good Charts)

[1] Scott Berinato, Good Charts: The HBR Guide to Making Smarter, More Persuasive Data Visualizations (Boston, MA: Harvard Business Review Press, 2016).

Click here to get the book!!


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

search previous next tag category expand menu location phone mail time cart zoom edit close