Picking the right color for your data viz is an important design choice. The data, the questions, and even the subject can all influence the color choices. Matt shares some tactics for getting the right colors every time.
- Colors in data visualization is always a tricky subject and that's also true when it comes to picking good colors for dashboards. Colors are very, very easy to get wrong. Typically, people use too much color. Tunnel makes it very easy to add color and it's so tempting just to throw as much color as possible However, too much color can be overwhelming and add to confusion. You should always ensure that the color comes from the data. Don't just add color for the sake of it. The color should enhance the data, make it easier to understand the data itself. Color can come from the questions that you were trying to answer maybe we just want to highlight the color rather than show every single possible value. It's important to not use colors just because you can, but color should always enhance and not distract. So look at the couple examples where the color definitely doesn't help understand the data and a simple fix that we can do to ensure that the color does help. So in this example, we have on the face of it, a reasonably looking color scheme. We can see that all the sheet look pretty uniform and overall the dashboard has quite a nice appearance because we've only got these same three colors being used everywhere. If we look at the individual sheets, we can see that we have our weekly categories so each line is colored by whether it's the consumer, corporate, or home office makes it nice and clear. Again in the customers we can see that we have the colors by the region, and for our sub-category profit, again, it's color coded by the category the furniture, office supplies, or technology. Now individually, those colors make perfect sense for each sheet. However, remember what dashboard is. A dashboard is not one single sheet. It's a collection of sheets in order to answer a specific question, or to provide a better view of that data set. In this example, when we apply all of those sheets together into a single dashboard, we get this which on the face of it looks fine. But, have a look at the legend on the top right. We can actually see the colors are different in each sheet. For the scatterplot, we are looking at the regions. For the weekly category profit, it's the categories. And for the sub-category, it's actually color coded by the segment, and not the category but in all three cases, exactly the same color scheme has been used, so what is blue in one sheet is not the same as it is in the other. In this example, what we've done is we've used the same default color legend for three dashboards, but it's encoding completely different information in each case. So on the face of it, although this dashboard looks quite nice and uniform in color and repeating the same color pattern, in actual fact it's very misleading because the colors don't mean the same on each sheet. Now a fix for this might be well, let's take away the categorical color and color it all by a value. Maybe, the profit. So if I go to the weekly category profit, and if I put the profit onto color instead of the segment, now each of the lines is color coded by the amount of profit that it made. I can then go and do the same thing for the customer sales and profit and I can do the same for the sub-category group right there now if we go back and look at our dashboard, it still looks nice because we have the same uniform color scheme and each sheet, the blue represents profit and the orange represents a loss. Now one thing to still bare in mind, if you look at the legends on the top right again, the maximum and the minimum value for each on those profits is very different because each of the individual sheets has a full range of color depending on what that sheet is showing. So those legends are still slightly confusing. So it might be a good idea to maybe get rid of those and just keep the color scale to show something is good or something is bad. So that's an example where we have the same color scheme, but the data is very, very different. Let's look at another example. Now this is an example where we are using a different color scheme on each individual sheet. Again, by themselves the sheets look fine. We've got three distinct colors that we've used for our categories. And for our customers, I've got four distinct colors so I can distinguish each of my individual regions, and again with the profit. However, when we put them all together in a single dashboard, it looks a real mess. So again, the fix for this could be to change them all to use the single color palette as we did before. Or maybe, we don't color them at all. What's the most important thing that we want to get out of this dashboard? Maybe it's that an individual customer is the important one. Maybe we just highlight those. So we have to be very careful when it comes to color. We should always make sure that the color is enhancing the data, not distracting it and certainly not misleading. If the color doesn't help, then maybe we just stick with one uniform color throughout, and ensure that good labeling and good database practice is enough for somebody to understand that data.
- Determining the purpose of a dashboard
- Making dashboards for specific audiences
- Creating dashboards for mobile devices
- Setting up an informative dashboard
- Posting explorative data
- Sharing insights on persuasive dashboards
- Designing organized dashboard layouts
- Adding actions to a dashboard