Throughout my career, I have given countless design reviews: with clients, with coworkers, with fellow UX designers, with product owners, and with prospective users. Every so often, a familiar question comes up. “Did you consider using a pie chart there?” The people asking the question mean well. The pie chart (along with its cousin, the donut chart) is well known and popular, but I have dropped it from my data visualization library.
Many people have already written about why pie charts are ineffective (two great examples can be found here and here), so I will only offer a quick summary here.
- Humans have trouble comparing areas.
- Humans have trouble making angle judgments.
In short, humans have a difficult time understanding pie charts, unless they are very simple (2 categories with wildly different totals). The one thing a data visualization is supposed to do is make insights about the data simpler, and the pie chart fails at this task in most cases, and is worse than alternate data visualizations (bar charts) in any other. To be blunt, pie charts are the exact opposite of being human-centered.
This begs the question, why are they still so popular, even among human-centered designers? I don’t think a solid answer has been offered, and I think this needs to be studied in more depth, because the research is clear about how poorly they support user understanding.
I will take a guess, though. Based on the prevalence, I imagine that users like them and request them. They are simple to talk about and explain and they can provide a nice visual. They provide a break from bar and line charts which explain data well. If this is true, this is where we as designers need to understand our users, their goals, and the greater good. We need to design in consideration of both how the user must interpret their data, and how they feel about the system. Like most things in life, we need to find a balance between these two factors.
We should do our best to ensure that the users are able to understand and use what we give them. Anything that gets in the way of that must go. However, even if the system is designed to perfectly match the mental model of the user (or of the physics of the domain – that’s for another discussion), if the user doesn’t enjoy the system, it will also struggle to gain traction among its users.
I’d like to be an optimist, and believe that pie charts are still in use because designers know the problems but use them only in low stakes situations, covering data that falls into only two categories, as a means to bring users a bit of joy. They choose to display important data in a format that better matches the decision-making needs of users.
The pessimist in me, however, fears that too much of UX is focused on what users want and not what they need. They don’t respect the perceptual and cognitive needs of users, and focus mainly on the emotional needs of users – what users like and ask for.
I like to compare this to children asking for candy. The parents know that the children should be eating healthy, but also know that the occasional piece of candy isn’t the end of the world, and will definitely keep the child happy. The goal is moderation and forming healthy eating habits in the long term. The same concept applies to pie charts. They aren’t good for users, but giving in every now and then isn’t the end of the world.
Can good human centered design contain pie charts? Yes, but use them sparingly and intentionally. Otherwise, you are doing a disservice to your users, no matter how much they ask for them.