Color Me Shocked: 5 Secrets to Improve Color Use in your Tableau Dashboards

Color is often cited as one of the hardest pieces to master in Tableau design. There’s a reason: conventional wisdom and trainings fail to be explicit enough. Believe it or not, there is a way to use color effectively without all the headaches.

Tableau has already done a ton of color science work, from creating color palettes that are balanced to developing diverging palettes that work with our ability to discern color variance. It’s not for lack of color…it’s often that’s there’s too much of it. Worse, many of us have been trained to think about color all wrong.

Challenge #1: Where We Color

Too often, we start coloring at the chart level (worksheets) right as we analyze. Even Tableau perpetuates this in trainings. We’re trained to build a chart, finalize it to a level, and move to the next chart. Eventually, we culminate this all to a dashboard. Different people spend varying degrees on how much effort they put into a single chart.

This has a downstream effect of requiring legends for just about everything. Legends are great on stand-alone charts. When you have multiples, it starts to feel like the office bulletin board where everything gets buried.

Removing color, once we have it, is hard. It’s like trying to pry off wallpaper and we struggle with it. Honestly, dealing with wallpaper is far easier.

Color tip #1: wait to color until you’re done building the dashboard.

Challenge 2: Universals of Colors

We’re trained to think color conveys more than does and that there are certain universals with color.

Wait for it…

Everything we know about color is culturally trained and learned. Our names and primary groupings of color vary by language, gender, and exposures. Language affects how our brain prioritizes the differences in color, in addition to physical differences in vision. As we create groups for things, we get explicit about the lines of differentiation and lazy about the middle. English speakers care a lot about the differences between L and R and B and P. A number of languages don’t and may group L/R as the same sound.

Which means most of our “universals” aren’t so universal at all. Instead, language imbues them with certain semantic meanings.

Does something about this dashboard feel off? You may have semantic associations with blue or red and find the conflicting uses here obnoxious.

Color tip #2: color may have semantic meaning, but it doesn’t mean it’s universal. (There’s more to this one.)

Challenge 3: Color is Everything

Design schools have most of us tied up in a knot about the importance of color. I give Pantone massive credit here in perhaps overstating color’s importance to design and creating a lifetime of work (not to mention a color of the year, so we go out and buy the same thing in 10 different colors. Don’t look at my closet or Andy Warhol experiments, thank you).

Here’s the rub – here are things color can do:

• Destroy the effectiveness of an otherwise well designed dashboard
• Invoke certain emotional responses (intentionally or unintentionally)
• Support or distract from the overall message

When we think about an effective visualization and task completion, we can do a lot without color. Here’s an example.

Without color, you can see all the other elements that comprise this dashboard. You get a sense how it’s weighted, how items flow to one another, and what the primary elements are telling you. It is – however – BORING. The end user can complete the desired task, but without some level of guidance, you’re going to lose them, so this is not the final product.

Color tip #3: Use monochrome design to get all other elements right. You’ll find a ton of other non-dataviz designers do this as well.

Challenge 4: Using Color “Wisely”

This is the crux of where we struggle. All advice says “use color wisely.” If we had that knowledge, we wouldn’t be hunting down this information. We end up in a vicious cycle of too much or too little and hating color.

Start in 1 color – usually grey, but it can be something else. See where you struggle to understand the message. Test with other users if you’re too close with it by letting them use it and seeing where they struggle with a given task. Where do they hover a little too long?

Color tip #4: Understand how your dashboard is used without color. Add in only what color helps to understand the message.

Challenge 5: Applying Semantics

Here I’ve added in color to the time of day which affects 2 charts. I’ve also done very subtle colors on the upper and lower hourly temperature bands. You can certainly play with color schemes all day.

My challenge is there’s a ton of blue and I risk overlapping semantic association between cold and time of day. I may love the colors, but find they hinder the message.

Color tip #5: Avoid overlapped encoding.

These 5 tips may help. There’s plenty more to discuss so there may be another round of color posts. Happy coloring!