Last week Bryan Briscoe and I delivered a presentation at the WorldatWork Total Rewards Conference in Washington D.C.. My portion of the presentation was about constructing visualizations that are “audience ready” – built to be understood quickly and thus maximizing your ability to influence. To illustrate a number of the principles I had identified, we shared a makeover of a visualization we had found to better utilize the key principles. I’m a big fan of the home renovation shows (in particular the Property Brothers), so this was my own little version of a renovation.
The audience feedback was tremendous, and thus the renovation is worth sharing here as well. We will do more of these periodically, since it appears to be helpful.
The original chart appeared in a 2009 article on the HBR blog. The premise of the article is about how the generational makeup of the workforce is changing (heard that one before?) and that the next generations – Millennials and “Gen 2020” will be very different.
Here is the chart:
To me, the big point of the article doesn’t jump out of the chart. Looking at the chart alone, I’m not sure I’d draw the conclusions the article is suggesting. As a result, I started down the path of searching out a more useful visualization.
An obvious first step is to see what Excel would recommend. Here was the first in its list:
It’s a happy-looking chart… and you get the sense that something is happening with the Millennials and the Boomers. But the key points still don’t jump off the page. The use of color is not very strategic, and grouping by generation rather than year just changes the complexity – it doesn’t remove it.
Excel also recommended this chart:
Just. Don’t. Do. It.
3D charts are misleading. Look at the Baby Boomers in 2015. The actual data point is 30%. Does it look like 30% to you? Excel aligns the top of the 3D bar with the plane that extends out into 3D space from the Y-axis, but on a 2D screen this doesn’t work well.
3D doesn’t add value, and detracts from clarity. Just say no.
When selecting a proper chart type, focus on what you are trying to say and highlight key phrases that suggest groupings, time, proportion, and focus.
In this case, our key statements are that the mix of the workforce is changing, and that Millennials are taking over. Note the highlighted words. “Mix” implies that we need to draw the reader’s eye to the composition of the workforce, not the raw percentages. “Changing” suggests that we need to keep an axis focus on time so that the change is clear. And the key point about Millennials would mean that we need to make sure the data about Millennials is n the spotlight.
Given these features, a more helpful starting point for the chart would be this approach:
Note how the stacked bars explicitly address the mix, and by maintaining the x-axis on time we more clearly illustrate how things are changing. So we start to have those covered. We can make the chart more aesthetically pleasing and use our space more wisely by making a few cleanups:
A few changes worth noting, and the reasoning behind those changes:
- I truncated the Y-Axis at 100%, since the mix cannot exceed 100%. While white space is helpful, that white space is not.
- I adjusted the colors to be more intentional. Grays represent the non-Millennials so that the blue shading for Millennials draws the eye
- I added the data points directly to the series. One challenge with stacked bars is that while you can see the proportion easily, you lose the ability to understand the absolute value. Because we want to make sure it’s clear that the Millennials are becoming the majority, it’s helpful to show how the value exceeds 50%.
- I added a chart title. Because you should.
After your first round of improvements, always step back and ask what other clutter can be removed. Here is my next version:
Some big important changes here:
- The previous chart included four time points… but why? If our point is about the change over 15 years, do we need to show the interim periods? Just because you have data, doesn’t mean you should use it. We can further de-clutter the exhibit by removing the interim years. This also makes the shift come across more dramatically, which is our intent.
- I removed the Y-Axis completely. Since the values are in the bars, the axis wasn’t adding much to understanding the data. This also meant I could remove the grid lines.
- I moved the labels from a lower legend to being next to the series, color coded the same as the series bar. This makes it easier to link the series to the generation.
A final step is always to make sure that the point is clear. In some cases, that may even mean adding the point in text. This isn’t always necessary, but for a written article it is helpful for those who only look at the pictures.
I’d love to hear your perspective on this, but I find the final version direct and to the point. The reader is immediately drawn to the Millennial data and the point is clearly made with less real estate than the original version. What do you think?