Chart Crimes: Multiple Pie Charts Edition

When it comes to displaying data in an informative way, there are good ways and bad ways to do it. Bad ways obscure our ability to understand what happened. Good ways showcase the most important points. Today’s Chart Crime is the multiple pie chart crime.

Recall that the purpose of a pie chart is to showcase the percentage of a whole. How do individual data points relate to each other? Pie charts are not good at showing relationships over time.

When I saw this chart of Lego colors, the tattered remains of my soul almost left my body.


To quote Anakin Skywalker, “Nooooooooooooooooooo!”

Again, pie charts don’t work well over time. As little circles, your eyes perceive them as discrete units. You can’t mentally carry data from one chart to the next beyond one or two pies.

Here’s a simplified version. Try to keep all three series in your head as you read these four pie charts:


Above, you can see the orange slice of the pie gets bigger and smaller, but the way the chart is laid out, it’s difficult to glean any usable analysis.

What should I have done instead? This:


A stacked bar chart accomplishes the same purpose as a pie chart in that it shows pieces of a whole and how they relate to each other, but it also does this well over time. In the earlier version of this chart, the orange slice grabbed your attention, but you probably didn’t notice as much the blue slice steadily growing. In the stacked bar chart above, that relationship is much more clear.

Any time you are tempted to make more than two pie charts, opt for a stacked bar chart instead. You will make life easier on yourself and make understandings easier on your audience.

If you enjoyed this, please share it with your network!

Want to read more like this from ? Get updates here:

Marketing Blue Belt Preorder

Order your 2016 Marketing Planning Framework

You probably don’t need a marketing dashboard

I love a good dashboard. The challenge of assembling one, of unifying data sources, of cleaning, transforming, and showcasing your data is fun. (This version of fun is why no one invites me to parties.)


Despite all this, most of the dashboards I have seen in my career are useless. In fact, they are worse than useless because the dashboard is an excuse, a substitute for the hard work we actually need to do.

Why? Decision makers don’t need data. They don’t need charts. They don’t need scatter plots with regression lines.

They need actionable answers to their questions.

What should we do?

What is the next step?

What is your recommendation?

What’s the plan?

When you hear these questions after you showcase your data, your dashboard, your analysis, it means you’ve fallen flat. It means that your work, hard though it was, ultimately didn’t achieve the goals that your decision makers wanted it to achieve.

Every analysis you do, every presentation you make must implicitly answer those questions above. Most of the time, a dashboard can’t actually do that. At best it’s a visual aid to your explanation. At worst it’s a distraction.

Before you launch a dashboard project or buy a dashboard tool, ask whether you need it to see that data for yourself or if it’s for your decision makers. If the latter, you probably don’t need a dashboard at all.

If you enjoyed this, please share it with your network!

Want to read more like this from ? Get updates here:

Marketing Blue Belt Preorder

Order your 2016 Marketing Planning Framework

Polls on Twitter Gone Wild

Marketers are abuzz with polls on Twitter. To quote The Flash’s Harrison Wells…

I do caution restraint

I do caution restraint.

Here’s why: Twitter polls are no more scientific or representative than sending out an email to your mailing list. Additionally, Twitter does not provide any kind of data which can be used for weighting.

Let’s look at an example. Suppose I ask my followers, “What is your opinion of @cspenn?”. What’s the likely outcome to be? Naturally, it will be skewed in my favor. Now suppose my arch-enemy runs the same poll. What’s the likely outcome to be? Of course it will be skewed in their favor. Which is the “right” answer? The answer is neither. Any individual account’s followers will automatically bias a poll.

This kind of bias shows up in many more places than just the obvious previous example. Here’s a sampling of Pew Research’s Twitter demographics:


Contrast this with the US Census Bureau’s Current Population Survey:


Note, for example, above that the representation of Black Americans on Twitter is more than double the actual population.

This sort of data skewing is problematic at best. If you’re running a poll to ask about an issue that’s of importance to race relations, you will get a different answer from the general population on Twitter than you would on, say, Facebook.

If you’re trying to ascertain the market viability of a product, if your product’s target audience is not represented on Twitter, you’re going to end up believing your product isn’t viable when it might well be.

Should you use Twitter polls?

If you want information about your specific followers’ perspectives, such as understanding why they follow you, or what their view is about you, Twitter polls are fine.

If you want representative, unbiased, statistically valid surveying of the general population, Twitter polls are not fine. They’re likely to dangerously mislead you. If you need that sort of surveying, hire a research firm to do it for you. You may pay more, but if your business is at stake, it’s worth the investment.

If you enjoyed this, please share it with your network!

Want to read more like this from ? Get updates here:

Marketing Blue Belt Preorder

Order your 2016 Marketing Planning Framework