Should I Use Pie Charts in Dashboards?
Are you an IBM Cognos developer? Are you tasked with rolling out reporting and dashboards to your end users but challenged to get them to use them? As IBM Cognos Business Intelligence experts and Business Analytics Coaches, Lodestar Solutions guides clients on how to gain recognition for building dashboards your users and executives will love. Today we will focus on, “Why you should not use pie charts.”
According to the BusinessInsider.com, “the pie chart is easily the worst way to convey information ever developed in the history of data visualization.” http://www.businessinsider.com/pie-charts-are-the-worst-2013-6
I am not sure I agree that it’s the worst. However, I have seen several charts that look like Rainbow Bright exploded and others that have almost sent me into an epileptic seizure. But Walter Hickey’s point is valid. You should not use pie charts because they are not the best at conveying information. Cognos Analytics, C11 BI, has so many amazing visualizations that will better serve the purpose than pie charts!
Let’s discuss why you should not use pie charts in dashboards.
First, your brain fills in the gaps. Scientific research has proven that in participants who had one eye covered with a patch, the brain fills in where there are blind spots. It acts as if it sees what it can’t see. The phenomenon called "filling in" has been known for some time. It happens when the brain "fills in" missing information in a person's blind spot if the other eye is covered. https://www.scientificamerican.com/article/brain-adapts-in-a-blink/ Similarly, when people are uncertain what the data is representing, the brain will fill in what it wants to see or believes to be true.
For example, what does this portfolio have more of: Stocks or 401K?
It’s hard to tell because we don’t have labels showing the percentages. But if we add the percentages, it can clutter up the graph. We have too many data points to add the percentages.
Let’s view the same information in a bar chart.
Now, it’s quite easy to identify that the 401K is larger than the stock holdings. Yes, the pie looks prettier, but we are here to provide actionable data and not confuse our users.
The moral of the story, whenever there is similar data, you should not use pie charts. You should not have to label each percentage for the user to use the chart.
Really, I should not use pie charts in dashboards, ever?
There are a few rare occasions where a pie chart may serve your purpose. For example, if the data set only contains two data points. A pie chart can be easily interpreted.
See how easy the above pie chart is to interpret?
The other thing a pie chart does have going for it is that it forces you to share a single measure.
I recently read an article by Stacey Harris VP Research and Analytics Sierra-Cedar, Survey Says: Upgrades and Updates are Still a Major Part of HR Technology Strategy. https://www.linkedin.com/pulse/survey-says-upgrades-updates-still-major-part-hr-strategy-harris and it contained insightful information. The article discussed the upgrade time for SaaS verses on premise software. (Yes, believe it or not, even SaaS requires upgrade time because you need to test it!) Stacey did a great job of sharing the information of the study and knows Why you should not use pie charts. My challenge however, was that the graph included in the article confused me at first glance.
I completely missed the fact that the top chart was in weeks and the bottom one was in months. I thought, “Wow a SaaS upgrade takes longer than on premise by about a 1/3. What?” Upon closer examination, I discovered that the two charts are using different time scales. My brain filled in and assumed the measures were the same. I just looked at the bars not the measures. Maybe I was having a blonde moment, not a senior moment of course, but, your users will have similar experiences. A way around this might have been to use different colors for the two charts or better yet, make the measures the same.
Moral of the story, you should have a standard measure to eliminate confusion. A pie chart forces a standard measure. But yes, you could make that same mistake putting two pie charts together that have different measures. We believe you should not use pie charts unless you only have two data points.
So, if now you are starting to question yourself, awesome! That’s the first step toward improvement! The second step is to learn from other’s mistakes by joining Lodestar Solutions for a real client’s dashboard teardown.
Join our webinar and witness a tear down and build-up of a dashboard. You will also see the new Cognos Analytics C11 BI in action.
It’s not your fault that your dashboards are not being used. You’ve been misguided in the principles of design. It’s time to learn what best practice myths are killing your dashboards end user satisfaction and adoption and register today for our free webinar.
If you need immediate assistance, check out our new a la carte services for small projects.
One of Lodestar Solutions’ Sr. Consultants could analyze your Cognos BI dashboards in a personalized, virtual tear down session. Your dashboard may be a few tweaks away from wowing your audience. For more information email firstname.lastname@example.org