A Day in the Life: Doing Good Design with Bad Colors

Posted by

Like anywhere, the place I work has good parts and bad parts. Generally, the good far outweighs the bad. I love my job, and wouldn’t get my medical care anywhere else in the area! And yet, I’ve never been a fan of our color scheme…

For a while, I just ignored the institutional colors. I’ve known for a while that this was something I needed to start doing, but I kept putting it off. Finally, after seeing enough great talks on corporate standards in Tableau, and hearing a sufficiently scary story (someone ‘drew’ over a Tableau graph in PowerPoint to get the right colors…), I decided to give it a crack.

This actually ended up being a refreshing challenge. I was in a ‘design dark period,’ and this made me think in a different way, breaking me out of my rut! And make no mistake, doing good design with bad colors is a challenge. But it can be a rewarding challenge.

Setting up the color palette
First things first, we needed some colors to work with. My coworker Tyler and I grabbed the official colors from the Marketing intranet site, but there were only two…
2 colors
Being limited to two bold colors isn’t great for data visualization. We needed more options, with different levels of color, so we set about extending these colors to have different gradations:
2 + light
I am a firm believer in the power of grey, and as such, grey got more options than yellow 🙂

Honestly, anything needing more than three shades of grey and two shades of an accent color probably isn’t best practice. But, we decided to add a few accent colors anyways. This isn’t strictly within the official institutional standard, but sometimes you need more than one accent color. Using a combination of free online tools (see below for links), we added four more accent colors, each with a lighter version:
more + light
The idea is to mostly use grey, using yellow for most accents, and to only use the other colors when truly necessary. 

A note on color deficiency
We did run this through a color deficiency (i.e. color blindness) simulator, and it did pretty well:
color blind quad inline

Red and green are always going to be a sticking point, but this palette seems to do relatively okay, while still having red and green actually look red and green. Our intended best practice is to either use red and blue, or green and purple – rarely/never using red and green together.

Putting it into practice
Once we had the palette set up, it was time to get to start using it!

I was already working on the following dashboard, so I started applying the new palette, and extended the ideas to fill out a full color scheme:
AYA main

As you can see, it’s mostly grey, which is by design. The yellow serves as an accent, highlighting the cumulative overall impact of the program being shown here.

A couple things worth noting…
1) I’m using Century Gothic instead of Tableau Book. While I love the Tableau fonts, they’re not installed in all web browsers, including the default browser at my institution. This makes your meticulously crafted dashboard look different than you planned. Others in my group were already using Century Gothic, as it works on all browsers (here at least). I liked the look of it, so I decided to go with it.

2) If you look very closely, you’ll see that there’s an extremely subtle background to the viz. Some people like extreme background colors, which I can appreciate. In a field like healthcare, however, I feel that it’s better to be understated. While the background color I used here is almost imperceptible, you can subconsciously feel it when you look at it. Plus, it helps to subtly, but explicitly, define the space of the dashboard. Here’s an example comparing an off-white background with no background (i.e. white):
white vs off-white

If you ask me, the light background is noticeably easier on the eyes than pure white! Small potatoes, perhaps, but that’s what design is all about: over-designing tiny details in order to provide a seemingly effortless and wonderful experience.

Plus, this is common practice when painting house interiors. No one uses pure white, we tortuously go through thousands of shades of off-white to find just the right one. This level of obsessiveness should be applied to data viz design as well!

Sequential and diverging palettes
I also developed sequential and diverging palettes based on the main colors. There is a sequential palette based on each of the six colors, and there are grey-yellow, blue-red, and green-purple diverging palettes. Here’s an example using the grey-yellow diverging palette:
Diverging palette

One more example
I’ll show one more example for now. I just like how this one turned out! It’s become a valuable tool for “Division A,” and I think it looks great:

Final thoughts
So, if you’re stuck with a color scheme you don’t like, instead of whining and avoiding the problem, think of it as a design challenge! Find a good grey or other neutral color that fits the company scheme, and use that almost always, using your accent colors sparingly and tastefully.

I’m honestly surprised how much I’ve come to love my new color scheme!

That’s all for now. Here are a bunch of links to check out if you’re thinking about making your own palette. Special shout out to Lisa Charlotte Rost for her wonderful blog post, which turned me on to many of these tools!


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

This site uses Akismet to reduce spam. Learn how your comment data is processed.