Learn the reason why to remove stuff (from Tufte). Find out why tables aren't always good or bad.
- [Instructor] Edward Tufte, my idol, and one of the undisputed masters of data visualization, you should really read his books, and he does not pay me to say that, says we have to remove all non-data ink. And I cannot think of a better way to say it. Non-data ink is any decoration. For example, showing pictures of beets in a chart reflecting beet juice sales. Also, that awesome wood texture, which I personally really like. This is an extreme example, but it shows the problem. A little decoration now and then is fine. Some audiences demand it, so we deliver.
But decoration is the bane of quality data visualization. No matter who your audience, you must remove all non-data ink from your data visualizations. If you learn nothing else in this entire course, understand this rule. Let's say my client sells beet juice, so I create this chart showing ROAS. I probably don't need to have the images of beets. This is non-data ink, so I'm going to start by deleting those. That's easy, right? So is removing three dimensions. You don't need a three-dimensional visualization to make this look the way you want it to.
Next, I'm going to remove the texture, because again, this is really decoration. I just need a solid color. We can even remove the legend, and if there are any tick marks, we can remove those, too. Now you have a cleaner, simpler interpretation of ROAS performance over time. It's still not beautiful, but it's much clearer. Our client can look at this and know right away whether we're doing our job, it is so much better. One last touch, by the way, if you're going to deliver this in Excel, consider filling in the table grid, just in solid white, like that, because it just removes one more thing that distracts my eye.
Now compare this to the chart that we started with. Which one does a better job of telling my client their change in ROAS over time? Don't start by adding stuff, start by removing it. Get rid of any representations that don't help communicate data. It completely transforms your data visualizations. It reduces mental tasks, and it improves the chances you'll be understood.
- Reducing mental tasks for your audience
- Removing redundant data ink
- Working with color and typography
- Creating a dashboard
- Highlighting data points
- Showing trends over time
- Determining which visualization to use and when
- Structuring reports