The Pros and Cons of Pie Charts (Mostly The Cons Though)

This is the first post in a multi-part series on data visualizations.

Okay, so just between us, I’m not a great fan of pie charts. I almost never use them. There are just so many better ways to display data and communicate information that I typically advise customers against them. That said, people love their pie charts, so I wanted to share a few thoughts as to “the good, the bad and the ugly” of the pie…

Why Pie Charts are Popular for Data Visualization

When it comes to data visualization, pie charts are like placebos: they occasionally have the desired effect, but they generally lack real substance.

So why does every dashboard have a pie chart? Why are they so popular? There are several reasons, including:

  • They’re easy to create
  • They’re easy to understand
  • They’re visually appealing
  • They’re named after a delicious dessert

Despite these perceived advantages, pie charts generally fail at doing what data visualizations are meant to do: make it easy to quickly see the relationships between segments of data. Report designers use charts and graphs because they want to communicate data without having to go to great lengths to explain what the individual numbers mean. They want their readers to quickly understand how one number relates to another, or how two series of numbers compare to each other, without requiring any additional background information. While pie charts are meaningful in certain narrow use cases, they are often far too simple to serve a purpose, and fail to provide useful insights when the data sets get even the least bit complex. As a result, they’re widely misused.

Below, we take a look at the rare cases when pie charts have some value, and the many cases where they should be avoided.

When to Use Pie Charts

There are a few situations where using pie charts can be effective.

First, when your data set is very small and doesn’t need to show progression over time, a pie chart might be an effective way of capturing it. For example, imagine you wanted to illustrate the share of overall sales revenue by quarter. You could take the overall sales number by quarter, and plot that into a pie graph with four segments. The result would be a clear illustration of which quarter saw the most success:

Sales by Quarter Pie Chart - Stytch Example

The challenge with this example is that the pie chart shows no linear representation of that quarterly data over time, so the reader is left guessing whether or not there was any growth in sales from one quarter to the next. Attempting to plot data that has some degree of time or linearity to it is a common pitfall in the application of pie charts.

A more appropriate application of a pie chart would be plotting those same sales by Product Group, providing there are no more than five or six Product Groups represented in your data:

Sales by Product Group Pie Chart - Stytch Example

Pie charts are also useful when you want to illustrate the dominance of one segment over the others. For example, if you want to show that Twitter drives more than 50 percent of all traffic to your website, dwarfing Facebook, LinkedIn and other sources of traffic, a pie chart might be an easy way to illustrate it:

Social Media Hits by Source Pie Chart - Stytch Example

When Not to Use Pie Charts

Businesses are now capturing more data than ever before, and as a result it’s rare that you’re only dealing with one data set. However, pie charts are not valuable visualizations when dealing with multiple data points. You just can’t plot multiple data points onto a single pie chart in any logical or helpful way. A pie chart will display only a single metric, such as “Sales” or “Qty Sold”. It cannot illustrate a relationship between these points.

Pie charts also don’t hold much weight when trying to compare data sets. We’re talking two big circles containing two different sets of data causing the reader to jump from one to the other, trying to make sense of things while slowly going crosseyed. In situations like these, a stacked bar chart could be a better fit.

In addition, pie charts are not useful when you’re trying to segment off one part of a data set. We’ve all seen it: a pie chart where the biggest piece of the pie has a big old “other” printed on it. If you’re looking at your top 5 countries in terms of sales, and you plot that on a pie chart, there are likely dozens of countries you’re leaving off the chart. As a result, the size of each segment can only be plotted in relation to the other segments. You lose the actual data behind the visualization and present a false view of the information you are trying to communicate to your reader:

Top Sales by Country Pie Chart - Stytch Example of a Bad Pie Chart

At the same time, leaving off the ‘other’ segment is equally problematic. The segments of a pie chart should always total 100 percent; if they fail to do so, the chart won’t only be confusing, it will be inaccurate. In other words, if you need the ‘other’ segment to get your point across, you should just use a better data visualization.

Finally, pie charts are essentially useless if you’re dealing with data points of a similar size. The whole point of a data visualization is to quickly convey a message. If the percent of overall sales contributed by each of your 6 sales reps is roughly equal (15 percent, 16 percent, 17 percent, etc.), and you decided to plot them on a pie chart, anyone looking at it would have an extremely tough time seeing the difference between these segments:

Sales by Sales Representative Pie Chart - Stytch Example of a Bad Pie Chart

On a column chart with an appropriate scale, the differences would be much clearer:

Sales by Sales Representative Column Chart - Stytch Example

Lastly, possibly the WORST use case for a pie chart is when analyzing data sets with large groups or population sets. In this case, the reader will never be able to determine the relationships between the groups or the percentages of the whole that you are trying to represent, as the slices will be so small as to be unreadable:

Sales by Country Pie Chart - Stytch Example of a Bad Pie Chart

5 Tips for Using Pie Charts

If you still decide that a pie chart is the right data visualization for what you’re trying to achieve, take a look at the following five tips before throwing that pie into your presentation:

  1. A pie chart with more than six individual segments is going to look very cluttered and lose its visual impact.
  2. Order your segments from largest to smallest. Don’t make people work to see the scale.
  3. Don’t try to make pie charts more visually appealing by adding effects like shadows or 3D perspectives. These actually make it more difficult to understand the data.
  4. Make sure everything adds up to 100 percent.
  5. Put a name or number onto each segment of your pie chart. A surprisingly large portion of the population is colour blind, so the chart is meaningless to them without labels.

Bonus tip: Look for a better way to present your data. Despite their popularity, pie charts are usually an ineffective means of getting your message across.

Want to dive deeper into data visualizations? Read our whitepaper on 6 Essential Chart Types for Any Data Visualization Need to learn more about the main chart types and how to choose the right one for your data.

Download Whitepaper

By on August 3, 2016

Leave a Reply

Your email address will not be published. Required fields are marked *

Demo

Copyright © 2016 Stytch Inc. All rights reserved unless otherwise stated.