(Download my improved Excel version of the bad charts show below.)
The Wall Street Journal usually publishes excellent charts. And Excel users usually can duplicate those professional-quality charts very closely in Excel.
However, in October, 2014, the WSJ published this figure, which offers several important lessons for Excel users. Unfortunately, the lessons are mostly about what NOT to do.
This figure, which is smaller than the original, of course, contains two rows of charts. The top row contains line charts, and there’s nothing much to say about them.
But the second row of chart figures is a disaster!
The light square in each figure represents the number of employees in 2003 for each company; the dark square shows employees for 2013.
The following figure shows that second row of chart figures in full size. To understand why the row of figures is such a disaster, think about why we use charts in the first place…
We use charts to help our readers understand numeric data as quickly and easily as possible. Designed properly, charts offer the closest way we have of transmitting business data directly to our readers’ brains.
But these charts fail to do that. The primary reason for this failure is that the human mind can’t compare areas easily. To illustrate, look at the red square above. When you glance at the chart—not the numbers—do you immediately notice that the dark square is about three times the area of the light square? I certainly don’t.
The WSJ editors must have agreed with this concern, because they felt it was necessary to plaster the actual number of employees on the face of each square. In short, the charts merely decorate the numbers shown; they’re junk charts!
Here’s my Excel version of this figure:
Here, I’ve used standard column charts to plot the number of employees for each company for the years shown. Because the human mind finds it easy to compare lengths of things, it’s easy to compare each column visually. And if more information is needed, the Y axis labels provide the approximate values for each plot.
This chart also uses the same colors for each company. Because the WSJ’s colors added no useful information, there was no reason to clutter the charts with random colors.
When I first understood what the WSJ’s figure was attempting to display, I wondered what percentage growth each company had experienced from 2003 to 2013. Because I thought that others might have the same thought, I added that information below each chart.
Notice that unlike the WSJ, my display does not show the actual number of employees for each company. Here’s why…
Most of the time, your readers aren’t interested in precise numbers in their reports. This is because precise numbers won’t improve their understanding of the data, and most readers can’t remember precise numbers anyway.
To illustrate, my chart shows you that IBM had slightly more than 300 thousand employees in 2003. The actual number, according to the WSJ, was about 319,000. Does knowing that number improve your understanding of the underlying relationships among the four companies shown? I doubt it.
Your readers are looking for insight, not documentation.
Going further, the original chart figure offers at least five lessons that you should keep in mind when you create dashboards or other chart-rich reports in Excel.
First, the purpose of a chart is to make numeric data faster and easier to consume. If your readers must study your chart to understand its format, or if they must sift through irrelevant data to find relevant relationships, then your chart has failed.
Second, companies that create charting software—like the program that created those square charts—do NOT create new chart types because their customers demand them. Instead they tend to create new chart types because their marketing departments keep begging for new features to talk about. That’s why BI Software companies offer so many types of gauges, 3D charts, radar charts, and other visual toys.
Third, most business data can be explained very clearly with common, two-dimensional chart types: lines, areas, columns, bars, and scatter (XY) plots.
Fourth, when you use common chart types, your chart still can have a very professional appearance.
Fifth, business publications—which can afford to keep graphic artists on staff—can give you some great design ideas for your Excel charts and tables. But don’t copy their presentation ideas blindly. Even great business publications can print some bad, bad charts!
