The Art of Data Visualization: Creating Meaningful, Influential Data Representation
The latest cutting-edge software is only as good as the visualizer behind it. With the right approach, visualization can be a powerful tool for telling a story and generating action. Data visualization can influence decision-making in all aspects of business, from prioritizing resources, to finding growth opportunities, to disseminating industry expertise.
When I present on data visualization, the first question I’m typically asked focuses on the technology (i.e., software) that I am using to create visualizations. The audience is often surprised to learn that the practice of data visualization actually predates our current environment’s high-power interactive software and technically challenging “big data.” One of the most influential discourses on data visualization, Edward Tufte’s The Visual Display of Quantitative Information, was published 35 years ago and offers many approaches that are still widely used today in the practice of visualization.
Readily available, sophisticated, and easy-to-use tools have democratized the translation of data sets into charts, graphs and maps. Many of the users of these tools are unaware of the core fundamentals of visualization. Successful visualization helps the viewer better understand a narrative built on data — unsuccessful visualization monopolizes valuable real estate in a presentation and,even worse, it can dilute or confuse the key messaging. The three most common data visualization mistakes are misrepresentation, over-complication and the “so what?” factor. To avoid these pitfalls, the novice visualizer should follow a basic checklist:
• So what? Start with this question to determine whether you really need visualization. Consider whether the visual conveys your point more powerfully than words alone and whether including a visual is truly necessary or just an attempt to make your analysis seem more technical than it really is. The goal of the visual should be to generate an insight or actionable response. You want the viewer to come away with a better understanding of the topic or to feel motivated to take action.
• Keep it simple. Once your visual passes the first test,the next step is to identify the most important message you intend to convey. Does your visual really highlight that key point? Consider using color and scale to spotlight your key statistics. Resist the impulse to overcomplicate your visuals with distracting tangential information. For exhibits with multi-dimensional data, solicit feedback from someone less familiar with your topic, and determine the length of time most viewers will need to decipher the meaning of your visual. At Sun Life, our visual development process includes a feedback loop with a pilot audience similar to our end consumer. If our test viewer requires a lengthy explanation to interpret the visual or gives up before finding the key message, we return to the drawing board and iteratively simplify until the design is clear.
For example, we recently updated the data visualization display in an analysis of business growth we provide to our key insurance broker partners. Based on focus group feedback, we decided to alter the representation of the benchmark, which better illustrated our most important message: highlighting the best products for cross-sell growth opportunities.(See figures 1.1 – 1.3)
• Keep it accurate. Afterconfirming that the visual is meaningful and intelligible, ensure that it accurately represents the magnitude and scope of the underlying information. Verify that the relative scale of your graphics convey the true relativity within the data. For example, the slope of a trend line can easily mislead as to the size of a change. If your data has a level of uncertainty, moderate the precision of the numbers that you display or include a bracketed range. Visualization should illuminate and not misrepresent.
Ultimately, the approach to creating a data visual is as important as the data itself. Well-constructed visual representation can turn impenetrable data into a highly influential message for your audience. Developing best practices for your team will help ensure that your visuals represent the data in an efficient and meaningful way.
The evolution of a graphic
The intent of this chart is to illustrate an insurance broker’s opportunities for cross-selling multiple products to existing clients. The largest opportunity lies with clients that have only one product. Four or more products is considered the gold standard.
Crowded. In the initial version of the graphic, the benchmark is illustrated as a vertical black line. The opportunity for cross-sell appears in the four+ products category, represented as the space in the grey bar to the left of the black benchmark line. In this case, the Small Client segment has the biggest opportunity for cross-sell only 7 of the 53 Small Clients have purchased four+ products. However, our focus group struggled to see the opportunity for cross-sell in each group because the chart does not highlight it clearly.
Better. A second attempt at the chart retains only the black line benchmark, adds orange shading and clearly defines each category relative to the benchmark line. This helped our viewers find the cross-sell opportunity a bit more easily.
Best. In the final draft, adding a text call-out for the cross-sell opportunity at the bottom and separating the four+ product data with clear, white space from the other categories really improved the visibility of the cross-sell opportunity for our focus group.