Improve Your Slide’s Charts
Types of slides that we often create are those that try to present data to our audience. All too often, this data is placed in a cluttered data table or worse, is presented in a narrative format. In most cases, a chart of some sort is the appropriate method of presenting data. However, there is an art to doing it right. A poorly formatted chart can be just as bad as that useless narrative on the screen.
In my experience, I've seen far more bad charts than effective ones. Often, the reason is expedience or because someone decided "hey, this looks cool!" An exotic sports car with a broken transmission may look very cool, but it's not going to get anywhere. Likewise, a poorly thought-out chart may look cool, but it's useless without an explanation.
We've got 3 seconds for that chart to convey its information in support of our story.
When creating a chart, the first thing we need to ask ourselves is, "what is the most important information?" The key words there are "most important" and "information." Information and data are not the same. Data are the raw numbers and figures. There's nothing more to them. Information is the message or story being told by the data. Data by itself will not translate into information. What we do to the data will provide us with information.
Secondly, we need to focus on which information is most important. Depending on how you present your data, the information will be different. I'm sure you've heard the phrase, "Lies, lies, and damned statistics." Well, it's true. How you manipulate or present data affects the message being presented. Let's look at two charts that contain the same data, but because of a different presentation, tell a different story.
For this example, I took the number of level 80 characters is one of my World of Warcraft guilds and created two different charts. The data is exactly the same in each. In the pie chart, the data is represented as percentages of the total while in the bar chart, the data is represented by the raw numbers. The message contained within both, while truthful, are misleading. In the pie chart, you have a large number of slices that well crammed into one chart minimizes the difference between them. Though you can interpret correctly that there are twice as many druids as there are warlocks, the visual and mental effort required to get there is more than is necessary. Without the labels, all we can do is guess at the relative values. In addition, since all we see are percentages (fractions of one whole pie), we lack any information about the size of the guild.
The bar chart tells a completely different story. Here, you've got a more apparent representation of the relative class populations. Is the chart poorly formatted? You bet. I'll get to that in a moment. However, in this example, it's immediately apparent that the warlock population is very small. A simple glance and the two bars tells that story in 3 seconds or less.
Bar Charts
Now, with that said, the relative populations of each of the classes is not what my most important information was. The information I wanted to convey was the population of death knights, added to World of Warcraft this past November, relative to to the other classes. The other bar chart example, this time from PowerPoint also fails to convey this information. In both the Keynote and PowerPoint examples, the chart settings were left mostly to their defaults. Below are the major problems with these charts:
- 3D In this example, PowerPoint's default is to present a 3D chart. The problem with 3D bar charts is that they lie to us. They give the bars more surface area than is necessary, and which is often out of proportion with reality, especially for the smaller bars. Take for example, the same chart displaying data for one of my other guilds. This chart has two bars with a value of 0, yet they have surface area. Thus, visually, these two bars appear to have value when in fact they don't.
- Inappropriate Title These charts are titled, "Classes - D'Gentlemen." Though it is true that these charts are displaying the number of level 80 characters in D'Gentlemen, sorted by class, that's not the most important information. What's most important is the number of death knights relative to the other classes.
- Bar Colors Obscure the Message When displaying multiple data points, it's not necessary to always give them their own color. Remember, what's most important in these charts isn't each bar relative to the other 9. What's most important is the value of one specific bar relative to the other 9.
- Sort Order Obscures the Message When displaying multiple data points, be mindful of the sort order. Sorting them alphabetically doesn't always make sense. Reordering the bars by value (alphabetically as a secondary sort) drives home to the audience the relative population of death knights in a manner that takes very little visual effort. The eyes don't have to bounce all around the chart to put all the pieces together.
- Chart Junk Edward Tufte refers to the stuff on a chart that doesn't add value to the chart "chart junk." In this case, the border of the chart, the grid lines, legends, and the numbers down the left all contribute to a data:ink ratio that is out of whack. None of that contributes to the easy absorption of the data. Because of limitations of presentation slide software and the nature of the data, our best option here is to place the value of the data points right there on the bars.
Once we clean up the problem areas of these charts, we're presented with two completely different representations of the same exact data. However, this time, the information that is most important is what is conveyed to the audience. The title was changed to "Death Knights - D'Gentlemen," the data points were reordered and all of the bars, with the exception of "Death Knight," are the same color.
Pie Charts
Another popular chart type is the pie chart. One key thing to remember with these charts is the information conveyed is always the data point as a percentage of the whole. Even if the labels show raw numbers instead of percentages, the message that your audience receives is that of a percentage or fraction.
My two examples of the before versions of PowerPoint and Keynote slides are of the 3D variety. Though not the default, they are by far the most popular. These charts are intended to convey the population distribution among the different races in my World of Warcraft guild. Let's look at where these charts fail us:
- 3D It can't be said enough that 3D charts lie. In the Keynote example, the 10% slices look just a little bigger than they should, all because of the illusion of perspective. In the PowerPoint example, because the gray slice is just as tall as the green slice, the gray slice appears to be bigger. This is because with a 3D pie laid out as it is, our brain expects the gray slice to be shorter. The rules of perspective dictate that identical height objects at different distances appear to be different heights, with the one furthest away being the smaller. You also have the same problem you have with bar charts where smaller data points get extra surface area than they should get, proportionately.
- Texture and Colors The Keynote example is using its default fill texture. Marble doesn't add any value and instead distracts the eyes with a lot of visual noise. Many extraneous lines are added when only a few, those that separate the pie slices, are important. The PowerPoint slide uses colors that just don't contrast enough and lets slices "hide."
- Lack Of Values Leads To Guessing The lack of data values in the PowerPoint example leads to guessing about the relative sizes of the slices. The legends in both leave a lot to be desired, but Keynote can move that information to the slices without compromising too much (depending on the slice sizes). PowerPoint can move this information to the slices as well, but the size of the chart is sacrificed too much.
- Chart Junk The PowerPoint example contains the usual chart junk: chart borders and slice outlines.
I cleaned up the charts by flattening the charts, simplifying the colors & textures, using more dense and informational labels (where I could), and eliminating chart junk. The new slides should be more effective when presented to an audience.














