Storytelling can be hard. Not the stories about dragons and lords and some kind of king of the nighttime, or stories about a dude on a quest to restore his soiled rug (it really tied the room together). No, the hard stories are the ones whose characters are simple numbers. Telling a clear, authoritative, and persuasive story using only numbers and data is something we analysts wrestle with every day. Sometimes, we tell a pretty interesting tale. Other times, by following a few guidelines, we knock it out of the park.
Over the years I’ve been on both sides of the table when it comes to telling stories with data. Unfortunately, I’ve seen myself and many analysts compile truly impactful research, only to fall victim to the perils of data overload, analysis paralysis, or, worst of all, potentially deceptive data visualizations. If you find yourself sitting on a great data set but are hitting some major creative question marks, here are some things to think about when trying to tell your story.
Who is your data being presented to? Are they expected to have a high level of exposure to industry terms and acronyms? Do they want the executive summary version or are they going to want to dig through the appendix and vet your sources? If you want your story to really be persuasive, start with your main thesis and think about the most efficient way to demonstrate what your audience cares about. Before getting started, here are two main questions to ask:
If you present a graph of your findings and the audience doesn’t recognize half the labels, you run the risk of losing them entirely. If you would have to explain every label, use different labels.
Worse than confusing or unclear labels, is no labels at all. This next graph is an example of using multiple axes but labeling neither of them. As a result, this graph fails on a few different levels. Not only is it confusing the actual performance numbers, it could very well be perceived as misleading.
Do we really suddenly have many more opportunities than we do leads?
Here’s an example of way, way too many dimensions. How could anyone possibly find the part of this graph they were looking for? Make sure when you’re building your visualizations, they don’t try to communicate too much at once. You’ll easily overwhelm your audience.
Avoid trying to add stylizing for the sake of being interesting or different. A graph is not going to be visually entertaining. Sorry, it’s not a work of art (although many of us may feel differently). The data that the graph communicates, is the interesting part. This is not a medium to sacrifice function for form.
This next graph is honestly pretty deceptive to the average casual reader and goes a long way to demonstrate how data can be clearly used but easily misunderstood. At first glance, this graph would appear to indicate gun violence in Florida has substantially worsened in recent years. Then after the passage of the Stand Your Ground legislation, gun violence immediately decreased.
However, the careful observer will notice the Y axis is inverted. The story of gun violence decreasing is a false narrative. The data shows that gun violence was actually lower before the passage of the legislation, and actually got substantially worse after the law was passed. To be clear, I have zero interest in the politics of this issue, just purely that the graph is misleading.
Then, of course, there are just plain nonsensical graphs. Pie graphs are generally used to represent which parts of a total population fall into specific categories, adding up to a total representation (100%) of the population distribution. The below graph goes for extra credit, with 193%! Also, which of the slices is bigger than the other?
Next time you’ve spent hours and hours pulling great insights from a complex data set, make sure you take the time to tell your story right. Make your analysis count by making sure your presentation is:
Analysts can be storytellers, too. And to quote Steve Jobs, “the most powerful person in the world is the storyteller.”
December 11 2017
October 10 2017
September 27 2017
September 12 2017
September 06 2017
August 30 2017
August 25 2017
March 31 2017
March 27 2017
© Method Savvy