How do You Communicate through Effective Data Visualization?
We have all been asked to convey data or other information in a visual sense – something that is easily digestible and does not need many words to get the point across. The first thing you might think of is to create some type of data visualization, or chart/graph. And, for the most part, people know the basics – line graphs show trends, bar charts compare categories, pie charts show pieces of a whole, etc. Once we think our visual looks polished – we share it and move on in our day. But when do we think about how the end user is going to digest this information? When do we think about the key takeaway we want end users to walk away with? When do we assess if we’ve been intentional with our message? We’ve taken courses on how to communicate effectively through language, but have we learned to communicate just as effectively through visuals?
De-Clutter, De-Clutter, and then De-Clutter
Have you ever looked at a data visualization before and glazed over it because it seemed like an onslaught of information – i.e., information overload? If you have, you are not alone. If a data visualization seems to take excessive energy to decipher, end users are more likely to glaze over and walk away without the information the visual was meant to convey. The solution? We can reduce the cognitive load it takes to glean insights from a visual simply by removing the clutter. Gestalt’s Principles of Visual Perception explain what our minds are naturally drawn towards:
- Proximity – We associate objects close together as belonging to part of a group
- Similarity – We tend to group objects with similar color, shape, size, etc. together
- Enclosure – We tend to assume objects that are enclosed together, either by a box or light shading, belong in the same group
- Closure – If there are gaps or breaks in a shape, our brains are able to fill them in despite the missing pieces
- Continuity – Similar to the principle of Closure, our eyes will always seek the smoothest path and naturally create continuity where it may not exist
- Connection – We associate objects that are physically connected as being part of a group
When a design is thoughtful, it tends to follow visual cues that inherently exist in our mind. Removing everything else that does not add to the interpretation of the information eases the cognitive load on the end user’s mind. As taken from multiple examples in Cole Nussbaumer Knfalic’s book and her website – these are a few steps one should take when thinking to declutter their visual:
- Remove your chart border – Focus on using white space to enclose your charts, which draws upon Gestalt’s Principle of Closure. With white space, your mind will naturally perceive that there is separation between the chart and other information but will do so with less visual clutter.
- Remove gridlines – Gridlines add to the clutter of charts and take away from the information. If your audience will need to manually drag their finger across the y-axis to identify a certain value, consider adding data labels to the points you want to highlight.
- (But) be sparing with data labels – Do not label every piece of information on your visual; instead, select the few you would like your audience to zero in on.
- De-clutter your axis labels – We are often in the habit of making vertical or slanted axis labels; instead, think about ways to reduce the characters like using abbreviation for month names or removing trailing zeros. If you have lengthy categorical labels you can’t shorten, consider placing them on the y-axis where it is easier to have horizontal labels. You may also consider having your y-axis title placed horizontally at the top of the chart rather than the vertically rotated title in the center of the chart. Make sure your x- and y-axis titles are descriptive and help end users easily interpret your visual.
- Move the legend directly next to the data it describes – Try to label the data directly in the visual instead of having a legend. This leverages Gestalt’s principle of Proximity and alleviates the mental load of having to refer back and forth to a legend.
- Leverage consistent color – If there are too many colors in a graph, it is hard to focus, and we lose sight of the main points. Try to pick one (or just a few) and use different tints and shades of those colors. Limit the use of your emphasis color to draw your audience’s attention to key points. You can also leverage Gestalt’s Principle of Similarity and make data labels or legends the same color as the data point they refer to.
- Add a takeaway title – How many times have you seen a chart with a title like, “Profits for X Company 2000 – 2015?” The title is probably the most important part of a visual and most often an afterthought. Instead of saying what your visual is in your title, consider having an action title that describes what you want your audience to take away from your visual. You might also consider employing subtitles to provide more context to your main title. If we title the same graph, “Profits greatly increased over the last 3 years,” our eyes are naturally drawn to the title first, by giving your viewer context upfront, they will understand your key takeaway faster and remember your main points longer.
Use Explanatory Data Visuals
Exploratory analysis is what you do to understand your data: it is the process someone takes to figure out (i.e., explore) what the most noteworthy takeaways are from the data. Explanatory analysis is when you have a specific piece of information about the data that you want to share (i.e., explain) with an audience. You will always have to conduct exploratory analysis to know which pieces of information are the most noteworthy for your explanatory analysis. But we often tend to hold on to all of the exploratory and explanatory steps we took to get to the key takeaways – and it shows in the visual. It becomes an exploratory visual, and you’re at risk of losing your audience by including all of these steps. By the time you get to the key takeaways you want to convey – you may not even have their attention.
Let’s use an example from Cole Nussbaumer Knaflic (storytellingwithdata.com): You work for a car manufacturer. You are presented with data on customer feedback to better understand how various metrics across multiple models impact customer satisfaction. This information is taken from a 5-point Likert scale survey (i.e., Very Dissatisfied, Dissatisfied, Neutral, Satisfied, and Very Satisfied). You’ve gone through the de-cluttering process and your first draft looks like this:
So, what’s wrong with this visual? It is certainly aesthetically appealing and conveys all of the relevant information in a reasonable manner. But were you able to easily conclude that the top two satisfaction issues had to deal with the steering wheel? What about in the following chart?
By highlighting our main point of interest and detracting attention from other elements, our audience can quickly focus in on our message. All of the other information is still there, so if the audience is curious, they may continue to read your visual – but they do not have to exert excessive energy in finding your main point.
Donut and Pie Charts are a No
For many reasons, donut and pie charts are not a best practice when visualizing data – especially 3D pie graphs. See Figure 1.
Notice how the “Other” category seems to be slightly smaller than the “Boston Celtics” category even though it is 7 percentage points greater. The “Boston Celtics” category seems to tower over the “Los Angeles/Minneapolis Lakers” even though they only differ by 2 percentage points. 3D pie charts – and pie charts in general – often skew the visual representation of parts of a whole, which may confuse your audience and put your audience at risk of misinterpreting or misremembering data. Skewing the visual representation also happens frequently in donut charts (Figure 2) because you are using arc lengths to visually compare areas and relative sizes of one category to another. Avoid donut charts at all costs.
Another thing to note is how many different colors we often find in pie and donut charts. This visual overstimulation detracts from the main point as it forces us to focus in on every single category. If you must use a pie chart, make sure to use colors sparingly – focusing instead on one color and its various gradations.
A good alternative to a pie chart is a simple bar graph – or a stacked horizontal bar graph that shows parts of the whole if it’s germane to the takeaway.
Tie your data visualization back to the audience
When communicating data visually, we want to ensure we both capture and sustain the audience’s attention…like a good story.
Aristotle introduced the three-act structure for plays that is still used to this day. In the first act, the main character is introduced and the setting is defined. The second act is the bulk of the story; it is where the main character is presented with a conflict and attempts to resolve that conflict. The third act includes the play’s climax – along with a resolution of the conflict and all of its subplots. We can take Aristotle’s structure for storytelling and tie it back to how we communicate our visuals:
- Introduction – Set up the context for your analysis. What are you comparing? What stake does your audience have in this? Grab your audience’s attention by asking them the questions they likely already have in their minds. Introduce the problem you are trying to solve to your audience.
- Tension – Address the problem you introduced and work with them to try to solve it. It is essential that you have a good understanding of who your audience is (e.g., their level of knowledge, the amount of time you have to present, etc.). Make it clear that they are in a unique position to make a decision or drive action.
- Conclusion – End your narrative with a call to action. What do you want your audience to get out of this? A good way to end a story is to tie it back to the introduction.
This narrative is one way your audience can easily remember and relate to the information because we’re hardwired to receive information this way.
Of course, there are cases where this narrative will change – sometimes you may need to present the raw data and the process it took to get there or you may need to lead with the ending to grab their attention. Adopt the mindset that sharing a visual in the context of a narrative/story is a good approach to take.
Implementing these four guidelines will help you more effectively communicate with visuals. These simple visualizations can have a huge impact on the success of your organization.
To learn more about how Sense Corp can help you with data visualization, visit our service page.
Sense Corp is a leading professional services firm transforming organizations for the digital era. We help clients solve their toughest challenges by bridging the gap between what is and what’s possible. To learn more about the services we offer, contact us.