In this article, I want to talk through three different charts that convey the same information. The purpose of this is to reinforce the importance that charts are tools of communication. My hope is that this article may help to promote thoughtful questions and considerations when navigating the interplay between visualizations and storytelling.

The charts that will be explored are bar charts, text boxes and pie charts as these are some of the most common visual aids used when communicating information. This article will explore some broad reasons why these charts are used and how they “work” (or don’t) with the given premise.

So, with that, what’s the premise? Using Super Store data, a user wants to know which subcategories are profitable and not profitable. Ideally, they should not have to spend too much time looking at their screen; instead they should be able to hover over a visualization to get more information.

Below are some important caveats to note before we dive into the charts.

First, for the purposes of this basic article, we will define profitability with this simple calculation shown below.

Second, every chart has the same tooltip configuration. An example is shown below.

Third, profitability is conveyed as “profitable” and “not profitable.”

Finally, the data was filtered to show only the top five subcategories by count, which means the charts will not showcase every subcategory. This was done to improve the legibility for each of the screenshots of the charts.

With the stage set, let’s take a look at each of our charts and talk about why they’re used and how well they “work” (or don’t) with the premise.

Let’s begin with the bar chart.

With a bar chart, our eyes are immediately drawn to the length of each bar and, in this case, how any given bar could be divided into two parts (profitable and unprofitable). We can clearly tell that there are more Binder and Paper based items sold than any other type of item shown. Overall, bar charts seem great for showcasing relative differences between sub-categories and work quite well for this premise.

While this next observation is outside of the scope of the premise and this article, it is still an important general consideration. When using bar charts, especially when conveying information on social ills or socioeconomic disparities between groups of people, bar charts can be overly simplistic and actually reinforce harmful and incorrect ideas and assumptions about groups of people and group differences. For more information on this, please read this article.


Next, the text box.

With the classic text table, we can see the actual values associated with each sub-category. If we wanted to, we could have hidden text labels for both the bar chart and the pie chart, which would make the text table the only visualization that displays the actual values of our data. Generally, however, text labels should be shown so long as the text does not overlap and the text is helpful and necessary. Not shown here in the text table, which usually are included are grand totals and subtotals. For the sake of consistency between the graphs, I didn’t include this as this could be visually jarring or difficult to implement for the bar chart and pie chart. Regardless, as can clearly be seen, there really isn’t much going with a text table other than it showcasing numbers and text.

A downside to the text table is that it takes quite some time, even with the color effect, to actually glean what sub-categories are most or least profitable or unprofitable. It is difficult to easily ascertain what are the highlights from the data. In this case, the ability to hover over some text and get more information using a tooltip is helpful, but doesn’t strengthen the usefulness of the text table for our purposes. So, maybe the text table is “helpful” in that it may give an impression that the data is correct as well as act as a “one-stop-shop” for being able to see what the data actually is. That is, there is no need to question the “value” of a portion of a bar chart or a slice of a pie chart. So, if a user just wants a clear and accessible list of their data, then the text table is appropriate for that. For more information about when to use text tables, consider reading through this article.


Finally, the pie charts.

Straight away we can see just how many pie charts there are, which may be a bit overwhelming and confusing. After all, given the current format of the pie charts, it seems that they are meant to do two comparisons: an “internal comparison” (is a sub-category profitable or unprofitable) and an “external comparison” (overall, is one sub-category more or less profitable than another). Personally, I think a pie chart should generally just do an internal comparison.

With that said, the pie charts generally do a solid job in visualizing the stark difference between profitability and unprofitability across the different sub-categories. In fact, I would say it does it better than the bar chart. However, it is visually odd that there’s a miniscule slice for the Paper sub-category pie chart. When using pie charts, besides considering how many pie charts to visualize at once, it is important to consider how many parts of the pie there will be. In this case, there are only two parts, representing profitability and unprofitability. However, in more “active” pie charts, you may easily see more than four parts, which makes it very difficult to ascertain what exactly is the difference between the various parts. To immediately see why this is an issue for communication and storytelling, take a look at this very active pie chart.

Having considered why these three charts are used and how they respond to the premise, we can further reflect on why it’s so important that we choose (and design) charts thoughtfully and purposefully. The charts that we design should be immediately useful and insightful for our target audience. They should help them to understand the larger story we’re trying to convey and give them the insights they need to make informed decisions. When we design our charts inappropriately, where we don’t practice data visualization best practices, we actively make the work of interpretation more difficult and may reinforce harmful and incorrect interpretations about the subjects we are visualizing. In all, we fail in our work as data communicators and waste the potential of data-driven insights.