Will somebody want to eat your pie?

Eating your pieI cannot understand why people are so fixated with pie charts. There are occasions when pie charts are useful, but here are more occasions when it is a completely inappropriate. I think many people decide on their chart, long before they actually decide on what they want to show…

Yesterday I discovered builtwith.com which allows you to see which technologies rae used to build any specific site. Just enter the web-site you are wondering about… Great tool for curios geeks like me! They also have a trend site trends.builtwith.com which provides interesting statistics.

However, this post is not about builtwith, but about using some of these graphs in presentations.

Maybe in this case we want to use the data to support our motivation for considering a specific technology. So I will use some data from builtwith. For the sake of explanation I am interested in payment providers, and specifically two questions:

  1. How is payment providers distributed between the top 10000 web-sites?
  2. Does the distribution change as one considers more web-sites, say 100000 or 1000000?

I both cases the builtwith website can help. (By the way the sell much more detail information at their Trends Pro site, which hopefully has more options to look at the statistics.)

Looking for an answer for the first question, I see the following graph.

Payment Technologies 1pie

Now this is fine for this question. Why do say so?

  • I can see Paypal is the majority, followed by CCBill.
  • Google Checkout is clearly in third place, followed by Flattr.
  • Mollie and Amazon Payments have no representation on the pie chart
  • They have a mouse-over facility which shows the actual percentages when you hover over it, but even without it (with this data at least) we can estimate the percentages quite accurately – in our presentation we may wat to show that inside the chart.

Pie charts are good at showing relative proportions, but horrendous at showing order when the data is close to one another. Consider, for example, the following pie chart showing the distribution of “Feed technology”.

Much more difficult to “see” the relative sizes, so unless you wanted to send the very basic message: RSS is leading, followed by 5 medium players; however, some smaller players still feature relatively strongly. So if that is all you want the audience to see, maybe good enough.

Back to the second question: “Does the distribution change as one considers more web-sites, say 100000 or 1000000?” If we then used the screen captures on our slides with a bit of cropping, cutting and pasting we could build something looking like:

Easily enough done, but can one really see the trends?  I can see that Paypal is getting much more popular as the web-sites popularity decreases… but for the rest I cannot really state anything.

Pie charts is no good at showing trends – and ever where it is visible it takes some extra mental effort to really see it. Now let is take that same data and represent it as bar chart.

I think it is much easier to see trends.

  • Paypal’s dominance is still easy to see, so we haven’t lost anything to the pie chart as far as I am concerned.
  • I suddenly saw that CCBill is get more popular as we move to the top sites. (Now that I’ve seen it, I can see it on the pie charts as well, but it certainly did not jump out at me in the pie chart)
  • I suddenly see some strange trend for Google Checkout in the middle… Not sure what it means, but it’s there…
  • Flattr also shows an interesting trend by appearing more in Top 10000 sites… than in other sites.

Did you see any of these trends easier with the pie charts? I know I could not.

So next time you think about presenting your message with data, please pay attention to what you want the audience to see, and make sure that that is what you show.


2 Responses to Will somebody want to eat your pie?

  1. ntari says:

    I would like to use pie chart for my research project but I am still thinking how I am going to use it

    • rabotha says:

      Mr Ntari, you are making the mistake of deciding on the data presentation before you have the data. When you have data you need to look at the data, decide what message the data gives you by analysing the data and only then can you decide which is the best way to show the data to another audience. There’s a saying that says you should not put the cart before the horses… that seems to be the case in your comment.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: