BI Tools

Tips and tricks for building information maps, OLAP cubes, reports, and dashboards

BI Admin

Learn your way around a SAS BI installation.

Visual Analytics

Learn your way around the SAS Visual Analytics tool

Coding & Data

Extract, transform, and load your data into the SAS BI toolset

Stored Processes

Create and design stored processes like a rock star

Home » Data Visualization, Visual Analytics

How NOT Knowing Pie Charts Makes You a #DataViz Rookie

Submitted by on 2015-05-07 – 10:15 AM 5 Comments

If I live to be 115 years old, I’ll never understand why consultants use pie charts. Your chance of doing them wrong is greater than doing them right. At the SAS Global Forum last week, I was delighted to hear several others telling me how much they hated pie charts and we agreed, they need to be stomped out – we are forming a #DataVizArmy to do just that.  Our first mission is to rid the world of poorly done pie charts.

Guidelines for Using Pie Charts

Here’s the guidelines for how to properly use a pie chart to display your data. I use the word guidelines because there are times (few and far between) that it makes sense not to do it this way.

Remember: Pie charts are used to show the parts to the whole.

  • Parts to a whole equals 100% – always
  • Limit to 4 or 5 categories
  • Doesn’t need a legend when done properly
  • Better when one category is significant percentage-wise
  • Not suitable for comparisons

When It Looks Good

Here’s examples of good pie charts. Notice that not having the percentage doesn’t matter that much – you know that Tracy is spending too much time trying to find the right content on Netflix and not enough time taking care of business.  (Names changed to protect the lazy.)

good pie chart

I like that the author went to the extra trouble of noting why the values may not equal 100%. Maybe having the percentage makes it seem more scientific?


Rookies Love Chart Junk

When you don’t understand the guidelines above, you are at greater risk for creating chart junk. Let’s look at some examples of when these guidelines were not followed to further convince you.

Is your data suitable for a pie chart?

Remember – the data is supposed to be the parts to a whole. What is the author’s point is with this pie chart? There’s not any numbers here – just dates and drug names. This chart only informs me that anti-coagulants were introduced in 1982 and more have entered the market since that date. It doesn’t express perhaps the different categories, the popularity, or anything really.  Use a line chart when you want to show a timeline or a simple line that shows the dates.

bad pie chart 01

Legends might make the pie chart superfluous

If you are thinking of adding a graphical element to the page, this is where you fall into the trap.  Really question yourself if you start reaching for a pie chart – that’s a rookie move.  For instance, does this chart really add anything besides some color?  I bet you just read the numbers and didn’t even notice the chart.


Famous example of Breaking the 100%

Again ask yourself does this data make sense as a pie chart? Just about anytime you read about pie charts – you are likely to see this example.  It’s hard to believe someone was so confused.  It’s back to that need to show data visually but not understanding what you are doing.  Are you glad they listed the percentage – otherwise the fault when have gone unnoticed.

bad pie charts 05

Limit the categories to focus the reader’s attention

When you have too many categories it makes it harder for the reader to understand your point. The reader may ask themselves “Is this a ranking?” or “Do these other categories really matter – why am I being shown this?”  Again notice how going back and forth between the colors and legend is a drag.  This would make more sense as horizontal bar chart.


It’s Hard to Compare Pies

Avoid using pie charts to compare to things – it’s harder on your reader.  The reader had to keep a lot of information in their visual memory – the order of the categories and the values.  So the comparison is really difficult and taxing.

pie charts hard to compare

Click for larger image

Review the difference when I rebuilt the chart in SAS Visual Analytics.  I used a butterfly chart, which is really just two bar charts back to back.  It’s a lot easier to compare the numbers and more clear how different the spending is.  Maybe you didn’t notice before that rich don’t spend a lot of their money on tobacco or perhaps it’s not a large enough percentage to register.  They do spend more on personal insurance and pensions!

Click image to see larger

Click image to see larger

More Rookie Alerts

What do you think the message is … “I don’t know what I’m doing?” or “Watch mom – I can do colors.”

bad pie charts


Never miss a BI Notes post!

Click here for free subscription. Once you subscribe you'll be asked to confirm your subscription through your email account. You email address is kept private and you can unsubscribe anytime.
The following two tabs change content below.

Tricia Aanderud

Director of Data Visualization at Zencos Consulting
Tricia Aanderud is a SAS Business Intelligence and Visual Analytics consultant based in Raleigh, NC who works for Zencos Consulting. She has written several books about SAS, presented papers at many SAS conferences, and has been using SAS since 2001. Contact her for assistance with your next project.
Spread the love



  • Tricia says:

    I agree – I usually prefer the horizontal bar charts when I want to show something.
    Of course you can also see I’m partial to the Butterfly charts – maybe I just like the name. 😉

  • Tricia says:

    I think that’s a good way to think of it.

  • LeRoy Bessler says:

    I don’t think of a pie chart or bar chart or line chart as an “analytics” tool,
    just a way to provide visual data insights more quickly and easily
    than a listing of numbers can.

    For the best way to present a large number of categories,
    whether significant or tiny,
    using SAS software,
    is with the horizontal bar charts presented in

    The bars are ranked,
    and the labels include Rank – Description – Value – Percent of the Grand Total.

    You use a macro which allows you to deliver one to four of any combination of the following subsets:

    – Top N (if smallness is the desired characteristic of the measure of interest, this could be Bottom N),
    your choice of N
    – CutOff (minimum or maximum),
    your choice of Goal to meet or exceed or Threshold to avoid crossing
    – Enough of the Top Values to Account For the Top P Percent of the Grand Total,
    your choice of P
    – All of the categories (not a subset)

    The macro delivers dynamically generated title lines with the following information:
    – Selection Criterion (per method 1, 2, or 3 above)
    – Observation Count for the Data Selected
    – Their Subtotal for the Measure of Interest
    – Their Percent Share of the Grand Total of the Measure of Interest
    – The Total Number of Observations Available
    – The Grand Total of the Measure of Interest

    If delivering more than one subset of the data, you can have the macro provide a title line of links to the other subsets.

    With the macro you can also provide an optional final title line for a comment and/or a listing of the Run Day Date Time.

  • Tom Kari says:

    Great post! I agree with everything you say. Generally, I view pie charts as “illustration”, not “analytics”.

    One place I like to see a large number of categories is when there’s a small number of “significant” categories, and a large number of “tiny” ones. As long as only the significant ones are labelled, it can be very illustrative.


  • LeRoy Bessler says:

    Let me respond to your guidelines for using pie charts, Tricia, restated here below, with my comment following each.

    TA: Parts to a whole equals 100% – always.
    LeRB: Yes, that’s the point of a pie chart.

    TA: Limit to 4 or 5 categories.
    LeRB: Only if there is some technical reason which makes handling more impossible, but offhand I cannot think of one.

    TA: Doesn’t need a legend when done properly.
    LeRB: When slices are many, a legend is likely to be a better solution than trying to label all of the slices. It eliminates the possibility of label-label overlap.

    TA: Better when one category is significant percentage-wise.
    LeRB: I agree only in the context of what I call “The Extremes of Other”. See my recent SAS Global Forum 2015 paper “Twelve Ways to Better Charts”, which can be found at

    TA: Not suitable for comparisons.
    LeRB: Eminently suitable for comparisons, when done right (again, see my paper referenced below). In fact, the whole point of a pie chart is to facilitate visual comparison of relative size of shares of the whole. That’s why people create them, and that’s why most people like them, AND understand them.