Color is powerful. We have a sense of this through our encounters in the world. The yellow traffic light catches our attention. We stare and admire the cyan water. We notice the vibrant green leaves of early spring. We don’t consciously think about it, but our perceptions are massively influenced by color. You don’t see many gray, dull advertisements in today’s day and light do you?
When it comes to analytics — presenting data with meaning and a message — color is a valuable tool. When used well, it elevates a visualization from ‘okay’ to ‘great’, ‘confusing’ to ‘understandable’, or ‘messy’ to ‘elegant’. While it’s tempting to throw in a lot of colors to make a visual “pop” or “make it pretty,” a color’s main use should be to focus the user’s attention where you want it to be. More color most certainly does not always equate to a better visualization. Line charts with 100 lines in 100 different colors don’t really help anyone.
Choosing the right colors for a Data Visualization depends on a variety of factors such as the goal (what you want to show, how it should look, and who will be using it) along with the subject matter of the Data Visualization, aesthetic principles, and accessibility informs a good color choice. Below we’ll explore several guiding principles of color use that can help you choose the right colors for your audience, data, and visualizations.
Your visualization is aiming to tell the audience something meaningful about the data so they can take the right course of action with what they learn. It is important to bring into focus the most important aspects of your visualization. One of the best ways to do this is through color.
As a general rule, it’s best to stick to one main color in Data Visualizations across the dashboard, while using gray or light black as a base color. Use color to call out something bad/low-performing, or use it for something good/high-performing. (Refer to the example below) The first graph forces you to search all the rows of data to process it, whereas the second graph highlights the low-performing products (less than 1000). It draws your eye and you comprehend which products are lagging behind more efficiently than the first graph.
Of course, some situations will require more than one color in Data Visualization. In those instances, a monochromatic color palette — one color but at varying levels of light and dark — can help cut down on the visual noise of the graph, especially if the color represents multiple categories. For example, if you want the color in Data Visualizations to represent a region, they can each be a different shade of blue instead of four different colors. See below.
When using color for categories, it’s a good goal to use five or fewer, or even think of ways to use color more effectively in your visualization.
Another way to make the color in Data Visualization more effective is to push the other data to the background. The “other data” is the data that isn’t the main point of the graph but still needs to be there. You do this by putting this data in gray. That naturally makes the main color you’re using that much more noticeable, but if the audience wants to look at the background data, it’s still there for them to explore. See the example below.
Even with a legend present, it’s very difficult to gain insight from the colorful spaghetti graph. The strategic, and minimal, use of color makes your visualization so much more powerful.
If you want to share the story of all of the lines and not only focus on one line, one solution is to make multiple charts but smaller, essentially mini-line charts divided up in a grid with each respective line highlighted and the rest grayed out.
Intentional Color Choice
As alluded to earlier, choosing to add more color just for the sake of aesthetics is not the wisest way to go. Color in Data Visualizations should be an intentional choice and not used almost as an afterthought. Colors hold a lot of connotations and guide our perception in ways that we may not realize. Our choice of color needs to be intentional.
Colors in Data Visualizations Carry Connotations
Colors like red and orange tend to connote an immediate attention-grabbing or something fast or intense. Blue and green, however, connote something “progressive” or complete or calm. It’s important to realize the implicit nature of this so we’re not encoding colors in a confusing way. We don’t want to color a profit KPI with red when it actually increased by 50% from last year.
There’s not a one-size-fits-all approach to color choice. Who your audience is, is vitally important to what you choose. For international audiences, the same color can mean different things than it does in your own culture. For example, green signifies good luck for Americans and Western Europeans, but red signifies good luck for the Chinese and Africans.
When using multiple visuals, whether in a presentation, in a dashboard, or multiple dashboards, keeping the same colors for the same dimensions is very helpful for the audience. For example, if you have three product categories represented in multiple visuals, maintain those same colors for the categories in all the visuals, and avoid using those same colors for other dimensions. If the colors changed throughout, it’s easy to automatically assume there was a change in the data. Color consistency allows the audience to familiarize themselves once — with the first visualization they see — and carry that understanding throughout the others.
There are a significant number of people in the world who are color blind, so we should aim to keep our colors accessible to them. As a general rule, it’s good to avoid putting red and green together. There are cases, however, where it makes sense to have those colors. For instance, in a KPIs context, it makes sense to use red to indicate a decrease and green to indicate an increase. Fortunately, there are a couple of ways to keep this accessible for color blindness too. One way is to double encode, in the red/green KPI example you can have a red down arrow for decreasing and a green up arrow for increasing. The arrow is another symbol for the performance that can be easily interpreted. Another way is to use different saturations of colors. A light green and a deep red color (or vice versa) is enough contrast to differentiate between the metrics.
A brief aside, a stoplight color palette (red, yellow, green) tends to be a popular choice, especially for KPI dashboards. But for the color blindness accessibility reason mentioned above, and for good design aesthetics it’s a good idea to avoid it. Those are loud, energetic colors and increase the cognitive load for the user to process. That is, it requires more brain power to make adequate sense of the visualization. A few suggestions might be to use a different color pairing, such as blue and orange to represent good and bad, and only add color where the data is “good” or “bad”.
Don’t settle for the default color options your tool gives you! There are a lot more options out there when it comes to color in your Data Visualization. Hopefully, this guide will help you make thoughtful color choices in the future. Use the power of color to elevate your visualizations, giving your audience focus and quicker realization of insights. Learn to use color in Data visualization with purpose and not let it work against you in the story you’re telling. Your audience will thank you, and probably keep asking you to deliver stellar visualizations.
I am indebted to Cole Knaflic and others for their expertise in focusing the audience’s attention. Color, as it turns out, is only one of a few ways to achieve this. Size, contrast, and position on the page are other preattentive attributes that strategically help focus the user’s attention. Check out Knaflic’s excellent book Storytelling with Data for more.
There are many resources out there that put you well on your way to effective color choice. The links below are an excellent start, but that is not to say you can’t identify opportunities to better utilize color on your own. Looking at the colors of the natural world gives us an ample amount of inspiration. Nature does its best, we would be remiss if we did not aim to mimic it.
Coolors.co – Color palette generator
Color.adobe.com – Color palette creator
Colorhunt.com – Color palette repository
Colors in Culture – visualization by David McCandless
Storytelling with Data by Cole Nussbaumer Knaflic