How to choose the best charts for your infographic

This article originally appeared on the Venngage blog. Venngage is a free online infographic maker. 

One of the most important steps in creating infographics is choosing the right charts to tell your story. How do you pick the best charts to represent your data in a unique and eye-catching way to successfully deliver your message? What are the techniques you can use to visualize your information so that your data speaks for itself? Here are some tried and true tips from the frontlines:

1. How to Convey a Single Important Number

Sometimes, you just want to convey a single data point. A very important number. Like breaking 100m world record in 9.2s. Or reaching the 1M users milestone on your new iPhone app.  Or winning a record 10th World Championship.

How to do tell this story? This is the time to employ the “Go Big or Go Home” technique. And by that I mean go big with your fonts or text.

Use Large Fonts and Labels

There is no need for fancy charts here. Plain old text works the best. Make your fonts and labels big. Make them stand out compared to the other text on the infographic.  Big numbers usually look impressive by themselves. Like Slack’s infographic above, the label is what draws your attention as it dominates over the line chart. Here are other examples:

Kobe Reaches 30000 points

To get a Wow reaction with one number or data point, it’s often better to tell your audience in plain large text. This is more effective than using a chart because you get right to the point and there is no interpretation needed.

Use large font to convey a single important number or point

 

 

 

Use Pictograms or Icon Charts to Complement Percentages

Although displaying large text works most of the time, having a visual to drive the point home can really help. In the case of an important number that can be conveyed in percentage or a ratio, a pictogram visualizes the number and really complements the text.

Similarly, you can use an icon chart – which is just an icon that has two different color tones. The percentage or height of one color represents the percentage or ratio. All of these charts are available in our infographic maker (it’s free!)

2. How to Compare Multiple Numbers or Things

The majority of data visualization deals with comparing a set of data points. There are many different charts you can use to show comparisons and which one you pick depends on your goal and what you want to emphasize. Let’s go over some of these options:

Use a Bar or Column Chart for basic comparisons

For 99% of the time, you can use a bar or column chart to show comparisons of different discrete things or categories. Most multiple choice survey answers can be visualized using a bar/column chart. Anything that can be counted and categorized can be easily visualized using a bar or column chart. For example, the total twitter followers of different hockey teams, the total sales of different products you sell, the number of windows in each room,..etc.

Use a Bubble Chart or a Treemap to Highlight Outliers

What if you want to emphasize a particular value that is an outlier? Or you have one value that is so much larger or smaller that you essentially want to compare it to all the other values in one visualization? You can use an area based chart such as a bubble chart or a treemap (which uses area to visualize the value as opposed to height or length as in the case of a bar or column chart). Using areas to visualize numbers brings out the contrast when there is a dominant category or value. This only works if there is an outlier, whose value is a lot larger than the others. Edward Tufte has famously dismissed 2D (area based) charts as an ineffective means to visualize data as we cannot discern area differences accurately. If all your values only vary by a little bit, then stick to the bar or column chart. But if you have an outlier, then visualizing the area differences really amplifies the contrast.

Here is an example of a famous infographic comparing disease deaths and funding that uses a bubble chart to show the contrast  of the how much we donate vs the actual diseases that kill us:

Here is another example using a modified treemap. In this case, it’s showing how big of a fine Pfizer has to pay compared to its profits.

3. How to Show Trends over Time

This one should be quite obvious to anyone who has seen a chart with a date on one axis. The most common chart to display time series data is the line chart. If you have many data points, the line chart is the most effective chart to show trends over time.

Use a column chart to emphasize distinct values

You can also use a column chart if there are not too many data points in the time series, especially if you want to emphasize the values (as opposed to the trending over time).

Special Case: Showing Periodical Changes

In business reports, it is common to show month over month (MOM) or year over year (YOY) changes. In fact, for many data visualizations absolute numbers are meaningless without the percentage of change or some other point of reference. How do you do a simple periodical comparison? First, you have to aggregate your data into months or years. Then you will need to calculate the MOM/YOY changes.

The table will look like this. The MOM% column is merely the change from the current month vs the previous month (the quick formula is current/previous -1)

The way I like to display this is to just overlay both charts on top of each other. In the example below, I’ve used a column chart to display the MOM% and a line chart to display the monthly widget sales. Here you can see that the month over month sales fluctuates drastically.

Special case: Show Non-Numerical Trends with a Visual Timeline

What about showing trends when you don’t have any numbers? You can simply display a timeline chart and add images for each time period.

Timeline charts are one of the most interesting charts around. To create one, you can use a vertical line to create a central column and horizontal lines as “ticks” for each time period. Then add your own text and images to each time period. Like this:

You can also create a horizontal timeline using a column chart as a base. Simply add all the time labels with 0 as values and turn on the Y axis. Then add your images and text:

Check out this article on how to create your own timeline infographic on Venngage.

4. How to Show Composition or Share of Total

How do you show the share of one value compared to all other values? There are 3 basic charts you can use:

Use a Pie Chart for most compositions

Despite its detractors, the pie chart or its cousin, the donut chart is one of the most recognizable charts today and can be easily interpreted by most people. Use labels and show percentages if you want to convey the actual values.

 

Or a Column/Bar Chart or Stacked Column/Bar Chart

The universally useful column and bar chart is perfect for this scenario. You can stack em or you not. In fact, not stacking them improves readability when the values are close to each other as you can compare the dimensions easily with a single glance. 

 

5. How to Display Rankings or Order?

The most simple way of displaying categories that are ranked or ordered is with a bar chart (with the categories sorted). For example, here’s the top 10 travel review sites:

You can also use a simple list or table or a combination of a list and a bar chart, like the example below for the top paying engineering majors:

6. How to Emphasize Patterns or Relationships

Most experts will tell you to use a scatterplot or a line chart or something more complex like a network chart or chord diagram to show relationships in your data. My take on this is that most people will find it really difficult to discover relationships themselves in any chart. Your audience will likely give your chart a two second glance, which isn’t enough time to figure out what you’re trying to say. You cannot expect your audience to interpret the relationships that your charts are conveying. Unless it’s really obvious, who can really recognize patterns looking at a hundreds or thousands of plot points? My unconventional advice is that the best way to convey a pattern or relationship between two variables is to just explain it in plain language, in a label or heading. It works a lot better.

Here’s an example of two line charts to show gender and credits earned and grade point average in math and science. Can you tell me what the relationship is between gender and performance in math and science? (Keep in mind, this example only has 4 data points. Most line/scatterplots have many more points)

Now look at the same chart with the relationships explained in plain English below. Is it much clearer? In fact, you should just display the 4 data points in a table and it wouldn’t matter. People will read the text and then look at the chart to reaffirm what they read, not the other way around. From my experience, it’s best to just tell your readers what the pattern is.

Remember, the whole point of data visualization is to deliver insights, not display data. If you can deliver an insight with a sentence or two, why not?

7. How to Show Distribution 

There are two general ways of showing distribution: Column chart (or a Histogram) and a Scatterplot.

Use a scatterplot when you have unaggregated data points.

If you have raw data such as exact time of all support calls, or the number of products a person buys during the year, or the NPS score from hundreds of clients, then a scatterplot to plot each data point is the best approach.

Use a column chart when the same data has been aggregated into groups or bins.

If you’ve taken the time to aggregate your raw data into groups, then a column chart (or a histogram) is a good way to showing distribution. How do you aggregate data into groups? With time based data, you pick an interval, say 3 hours, and lump all the data points into groups based the interval (ie, group 1 would be 12 -3am, group 2 – 3:01am -6am and so on). Similarly for values, you create groups for a range of values (i.e., 1-5, 6- 10, 11-15) and then add up the number of records that fit into each group. You’ll end up with a column chart like the one below:

As with the previous point – if there is a clear pattern in the distribution, you should write it down and not leave it to the reader to interpret. For example, most support calls occur late in the afternoon. You should add a label in plain language for your reader. The text complements the chart together to reemphasize the pattern.

Insight in plain text  + chart is far more effective than just displaying the chart

8. Avoid complex charts

Lastly, if you’re ever tempted to incorporate a very complex chart, or if the marketing agency or consultant you’ve hired tells you that you need a large and complex chart, stop. While large complex charts may look “cool” or beautiful, you should keep readability in mind. How many people can actually understand a chord diagram or a sunburst chart with thousands of data points?

Keep it simple. Summarize your data into a bite size that is easily readable. Stick to standard charts that a 6th grader will understand. (Incidentally, newspaper writers usually write at a 5th-7th grade literacy level.) Unless your audience are academics or researchers, our general recommendation is to use standard charts and to use labels to interpret the charts as much as possible. Rule of thumb – if you need to zoom in to look at a chart and its legends more than once to understand what’s going on, it’s probably too complicated.

Hopefully this was helpful. Send me a message if you have any questions.