Lessons learned from “Storytelling With Data: A Data Visualization Guide for Business Professionals”
When it comes to effective data visualization, the very first and also the most critical step is to select the right graph/visual for the data that you want to present. With a wide range of visualization software that is available offering a large number of chart varieties, it’s often a challenging task to pick the right one, which explains the data and insights in the simplest possible manner. I recently read a very famous book on data visualization — “Storytelling With Data: A Data Visualization Guide for Business Professionals” by Cole Nussbaumer Knaflic. This book is the best resource that I’ve seen till date on data visualization, and in this article, I’ll explain a topic from the book — Choosing an Effective Visual.
Most of the data can be visualized using any one of the 12 kinds of visuals that I’ll be discussing in this article. The visuals can be classified into:
- Simple Text
- Table (table, heatmap)
- Points (scatterplot)
- Line (line graph, slope graph)
- Bar (horizontal, vertical, stacked, waterfall)
- All the graphs shown are made using Google Sheets. Link to the document.
- Data used for generating graphs are entirely imaginary and not taken from any source.
So let’s start exploring each one on the list.
You don’t have to always use a graph for showing numbers. If there are just a few numbers with some supporting text, directly showing the numbers might be the best way out. Let’s look at an example to understand better.
In the above case, the graph doesn’t provide much aid in interpretation and only ends up occupying a lot of space. So, when you only have a few numbers, show them directly.
If you’re looking to communicate multiple units of measure, a table might be the right visual to use. Creating a table is pretty easy, but always make sure that the design fades into the background and data is the main focus. Here’s an example of fading the design to the background and focusing on the data:
Can you observe the improvement after every iteration? This is why it’s so important.
Heatmap is simply an upgraded version of a table where we add colors to interpret the data or numbers better. In a plain table, the reader has to scan every element to get a sense of what’s there. By adding colors, we are making the reader to directly focus on the area of interest, which results in a better understanding of data.
Graphing applications like Excel have conditional formatting options to create heatmaps. And it’s also a good practice to include a legend for better understanding.
Scatterplots are useful for showing relationships between 2 variables where each variable is encoded in X-axis and Y-axis, respectively. It’s especially useful while explaining correlations.
Line graphs are best when it comes to plotting continuous data like date and time. Since all the points are connected using a line, it’s easy to interpret continuous data, but at the same time, it doesn’t make sense for plotting categorical variables. Line graphs can be used to show a single series or multiple series of data, as shown in the figure.
Slope graph is simply a special case of line graph which is ideal for comparing change in metrics over two different points or time periods. This is really good to intuitively show the rate of change (increase or decrease rate is indicated by the slope of lines) along with the absolute values.
Next, we’ll look at a few variations of the bar chart, which is ideal for categorical variables. Bar charts tend to be avoided because they are common, but since they are common, it’s very easy for the rea
ders to understand bar charts compared to other types of visuals. This makes bar charts one of the most important forms of visuals.
This is the plain bar chart where each column represents a category. Similar to line graphs, bar charts can also hold multiple series.
Stacked Vertical Bar
Stacked bar charts can be used to compare subcomponent pieces across different categories. It can hold either actual numbers or percentages using a 100% stacked chart.
Again you mustn’t stuff the categories with too many subcomponents as it becomes difficult to understand and compare.
A waterfall chart is another special case of a vertical bar that can be used to either pull subcomponents of a stacked bar to focus one at a time, or to show a starting point, increases and decreases, and the resulting ending point.
A horizontal bar is often the go-to option for categorical data because it’s easy to read than the vertical bar and can also accommodate large category names. Similar to vertical bars, it can also have single or multiple series of data.
Stacked Horizontal Bar
This is similar to the stacked vertical bar chart but comparatively better because of the reasons discussed for the horizontal bar.
Area graphs should be avoided whenever possible because human eyes are not so good at comparing values in two-dimensional space. But if you badly want to include multiple metrics, then the area graph might work out.
With this, I’ve covered graphs that can be used to visualize a majority of data available out there. So choose a graph that can clearly explain the message that you’re trying to convey.
As we’ve gone through the best practices, now it’s time to look at some of the practices to be avoided.
Visual Practices to be Avoided
Avoid using pie charts because the readers have to compare areas of the arc, which becomes very difficult and is not intuitive. Using a standard bar chart makes it much easier to interpret. Look at the example below to understand better.
Never use 3D charts. 3D charts create unnecessary distractions and make it difficult to interpret. So never use 3D.
I hope this article would have given you a good understanding of different visuals and the right place to use each visual. So always choose a visual that adequately conveys the information you are looking to present. And, coming to the application/software that you can use, it’s entirely up to you. Excel, Tableau, Power BI, Google Sheets are some available applications, and you can use anything that you are comfortable with. Remember that the graphing application does not know the actual purpose of the visual, and it’s on you to customize it according to the need. I hope it helped.