Data visualization is a powerful tool to communicate complex and nuanced information in an engaging and memorable way. But how do you create a data visualization that tells a compelling story that resonates with your audience and drives action? In this article, you will learn some key principles and tips to craft a data visualization story that is clear, relevant, and impactful.
Before you start designing your data visualization, you need to understand who your audience is, what they care about, and what they need to know. This will help you tailor your message, tone, and style to suit their preferences and expectations. For example, if you are presenting to a technical audience, you might use more detailed and sophisticated charts and terminology, while if you are presenting to a general audience, you might use simpler and more accessible visuals and language.
Knowing your audience is probably the single most important component when presenting findings. Different audiences care about different insights and this should be taken into account when presenting. Data scientists and ML engineers might be more interested in model performance (AUC, R2, etc.) or clever feature engineering approaches. Business managers and product teams will probably care more about the ROI implications and revenue impact of improved ML models or exploratory data analyses. By knowing our audience we can better align the key takeaways to suit different needs and convey the right message.
Knowing your audience is important for picking an appropriate data density in graphics. When I'm presenting to a niche highly-technical audience who are familiar with the subject matter, I use denser graphics with more data points per slide, but when I'm presenting to a broad audience who are more interested in the high-level understanding, I try to strip down the graphics to only make one point at a time.
The format of your data visualization should match the purpose and context of your story. Depending on what you want to show and how you want to show it, you might use different types of charts, such as bar charts, line charts, pie charts, maps, or scatter plots. You might also use interactive or static elements, depending on how much control and exploration you want to give to your audience. For example, if you want to show a trend over time, you might use a line chart, while if you want to show a comparison of proportions, you might use a pie chart.
Designing data visualizations should follow some basic principles to ensure clarity, consistency, and aesthetics. Colors, shapes, and sizes should be used to highlight and contrast important information, but without being too many or flashy to distract or confuse your audience. Labels, titles, legends, and annotations should be used to provide context and explanation, but without cluttering or overlapping with the data. Scales, axes, and grids should be used to show the range and distribution of the data, but without misleading or inconsistent elements that might distort or misrepresent the data. White space, alignment, and balance should be used to create a visual hierarchy and harmony, but without too much or too little that might create gaps or crowding.
6. Symmetry: People tend to perceive symmetrical objects as more aesthetically pleasing and easier to remember. This principle can be used to create balance and harmony in data visualization by using symmetrical visual elements.
5. Figure-Ground: People tend to perceive visual elements as either being part of the foreground or the background. This principle can be used to create visual contrast in data visualization by using contrasting colors or textures to highlight important information.
4. Continuity: People tend to perceive smooth, continuous lines or patterns as more pleasing and easier to follow than jagged or interrupted lines. This principle can be used to create smooth transitions between visual elements in data visualization.
3. Closure: People tend to perceive incomplete objects as complete. This principle can be used to create visual elements that suggest completion, such as using lines to connect data points in a scatter plot.
2. Similarity: Objects that are similar in shape, color, or size tend to be perceived as a group. This principle can be used to create visual groups in data visualization by using consistent visual elements to represent similar data.
Data visualizations should not only present the data, but also tell a story that resonates with the audience’s interests, emotions, and actions. To do this, you should have a clear and concise main message that summarizes the main takeaway of your data visualization. Additionally, your data visualization should have a logical and coherent structure that guides the audience through it. Moreover, it should have a narrative and engaging tone that captures the audience's attention and curiosity. Lastly, you should strive for a relevant and impactful outcome that shows the implications or consequences of the data visualization for the audience or the world.
To ensure a compelling data visualization story, you should test and refine it based on feedback and evaluation. Ask your intended audience or a representative sample questions such as what they learned, liked or disliked, found clear or confusing, interesting or boring, relevant or irrelevant, and impactful or trivial about the data visualization. Then, use their responses to refine your data visualization and make it more clear, relevant, and impactful.
This is going to sound silly, but I like to get a fresh view on my data viz by literally moving around the room to look at it from a different angle. Stand 10 feet away, then really close to the screen, then print it out and look at it on paper, then go take a nap and look at it again. Does it still make sense?