Data visualization is a must for everyone doing any kind of research that produces long table sheets full of numbers. But, if done wrong, it can make things worse and instead of helping the audience understand the primary message the visual is trying to convey, it could tell an incorrect story and confuse the viewers.
There are 3 simple rules for visualizing data correctly:
- Highlight your message and eliminate distractions.
- Use visual cues to guide your audience through your insights.
- Use contrast (size, color) to capture the viewers’ attention.
Many times one or more of these rules are broken into a single data visualization.
Let’s take a look at this graph by Deloitte, which is a result of recent research about the number of Millennials who are expected to leave their current job in the following years. Several questions come to mind while looking at it:
- Where are the Millennials going?
- Are that 19% and 22% of Millennials especially important?
- How much is 27%+66%+8%? Is it 100%?
- Why do the colors divide the attention precisely in half?
As you can see, there is a lack of clarity in this visualization. Here is a short analysis:
Highlighting the message
At first, the message seems pretty clear. The one most important finding from the research is stated in bold at the top of the illustration (Figure 1. Two in three Millennials Expect to leave by 2020). However, looking merely at the figure and its title it’s not clear what are the Millennials leaving (at least not until you use the magnifier for the petite typed question at the bottom and connect the dots). On the plus side, the huge “66% expect to leave” mark resonates well with the message.
Usage of visual cues
A few visual cues in this graph are trying to help us where to look to find the most important message. Even though, when you first look at the chart, it takes some time to understand the way different groups of Millennials are divided and represented, you immediately notice that 66% who are leaving which is good because it’s the most important thing that this visualization is trying to say. The elongated 19% and 22% pieces of the half-pie suggest that this is the second most significant insight.
Capturing viewers’ attention through contrast
Contrast is surely used here, both in color and in size, but could be more properly applied. Two colors and various hues are used to express same type variables (both green and gray tones are used to represent a period) and then a shade of green is also applied to demonstrate the percentage of those that would never leave. This group of people should be represented with a third color. Moreover, the chart would work much better with reversed colors: green for the percentage of Millennials that would leave their job until 2020, gray (neutral) for those that wouldn’t. In that case, it would make sense to display the ones that would never leave with black.
Has the use of different sizes helped in demonstrating the real message here? Probably, but dividing the attention in half creates the illusion that both sides of the chart are equal (50% each), and that the right part is just more important. That is true but is also possibly the reason that the 19% and 22% slices had to become so chunky.
It’s clear that the main idea was to display the percentage of Millennials that would probably leave until a certain point in the future (that’s 2020), but that should’ve been done with a more thoughtful design and a clearer title. For example, we could use three groups of colors instead of two. Each color coded segment could represent a different group of variables (1. from 6 months to 5 years; 2. more than 5 years; 3. would never leave) while the ones that didn’t have answer could stay transparent, unfilled with color or whatever. Even reading the sub-header with the variable “would never leave” doesn’t sound okay (Percentage who expect to leave in the next… Would never leave), and it also shouldn’t have the same color. That being said, the title area and the legend should definitely work better together.
Bonus riddle: 27% of Millennials reported to stay, 66% will probably leave until 2020, and 8% don’t know. So, do the math: Did you get 100%?
There are many examples of poor data visualization on the internet. Let’s see a few more.
1. Why is the blue dot the same as the one from the USA Today logo? What are those arrows doing? What is DC doing zoomed in over there?
2. Why would someone combine costs with math scores? Some metrics just don’t fit in the same chart.
3. Why overcomplicating things?
4. Did someone have trouble aligning the percentages with a proportional surface? It seems like the tip of the thumb doesn’t provide enough space for 88%.
5. Is this suppose to be a joke? Where’s the red in the legend? What does the break in ‘design’ mean? What are these folks doing with the rest of their time?
We can’t really learn how to do data visualization correctly, but following the 3 main rules to visualizing insights with impact is a good start. Also, we should always ask ourselves what do we want to accomplish with our visualizations and do several tests before we finalize them and present them to the audience. Here are a few that help a lot:
- Look at the visualization and ask yourself: Would eliminating any of this change anything? If the answer is ‘No,’ get rid of it.
- Look away from the visual for 5 seconds, then back on it. Where is your eye drawn? Is the first thing you see your most important message?
- The colleague test: Ask a co-worker who doesn’t understand what you do to take a look at your visualization for 10-15 seconds. Give them just a general idea. Then, ask them what they’d take away.
If a flawed chart is viewed by the important people in an organization, someone could get in real trouble, and if we confidently decide to share our ‘good’ work online, we would probably end up being mentioned here.