Good Charts: The Book I Wish I’d Written

It isn’t often that a book comes out and I say “I wish I’d written that book.” But if I could have double-clicked on the data section of slide:ology and had a book pop out, Good Charts would be the book! It’s a remarkable trek through how data should be explored and displayed—and not just for data scientists and designers, but for everyone. The book’s author Scott Berinato argues that data visualization will soon become an essential workplace skill and that those who learn how to do it well will be the ones who get noticed and contribute to their companies’ success.

Berinato, also a Senior Editor at Harvard Business Review, was gracious enough to stop by Duarte’s offices recently and nerd out with me about data viz. Here’s an excerpt from our conversation.

Nancy: One of the things I love when you open the book up is you have a very simple chart on the dedication page. What does this chart mean and who is the book dedicated to?

Scott: S, E, and M are my wife and two daughters, Sara, Emily, and Molly. They’re off the charts. Very simple, but if I’m doing a book on visualization, I might as well visualize the dedication.

Nancy: Where does your data geek side come from because most editors aren’t data geeks?

Scott: I’m not a data scientist by any means. I can mess with data, but I’m not a professional designer as much as I wish. But I’m a visual learner. It turns out when I started researching it, we’re all actually visual learners. Vision, in a lot of ways, is what the mind does. I became obsessed with it about 10 years ago. I was researching a story on risk, and I’m fascinated by people’s inability to understand risk. It’s really hard to understand things like 40% of chances of something happening or 10% chance or 7 parts per billion. These are abstract ideas. Visualization is one of the primary tools we have to deal with probability. It’s sort of grown in interest for me over time.

Nancy: There are a lot of clever ideas in your book. One of them is the two questions you can ask yourself so you know what kind of chart to be making. Can you talk about that?

Scott: Sure, I asked myself how I would organize the different types of visual communication we do. And there are really just two questions: Is it a conceptual or a data-driven idea? A data driven idea is obviously, it’s statistics, right? Revenue, sales, whatever. Conceptual ones are anything that you want to visualize that doesn’t have numbers behind it. And we do a lot of that in presentations, especially, like process diagrams.

And then the other question I wanted to ask is: Are you saying something, declaring something, or are you exploring something? Because even within data visualization, the difference between a chart that says, “Here’s our revenues,” and a chart that you build, which you want to learn what happens to our revenues on rainy days, these are completely different activities. They require different skills and different techniques. So, we wanted to break it up into these four pieces.

Nancy: And what are the four pieces?

You have data-driven declarative charts. That’s what you present to the board. That’s your classic bar chart, line chart. You have declarative conceptual charts. Those are your consultants’ diagrams, your process diagrams, your org charts, that kind of thing. You have conceptual exploratory; that’s your whiteboard session. You say, “We want to reorganize our business, and we’re not sure how it should be structured. Let’s start drawing to see what shape the org may take.” That’s more of a brainstorm. And then you have sort of the most exciting for me is the data-driven exploratory category, which is a lot of what’s happening in data science, but managers are getting into it now, too, where you don’t really know what you’re going to show yet. You’re looking for patterns in the noise. You have all this data, and you start forming hypotheses and testing them and seeing what comes out.

Nancy: In the book, you show the same data multiple times—the same exact data—but it took on different types of meaning based on how you visualized it. You did a really good job of keeping the data pure and showing the difference between what it means to display it different ways.

Scott: I think as the data gets bigger and as the ability to create views into that data get more and more multiple, you’re going to see more and more people pushing the limits. And we talk a lot about the ethical considerations there. But I do say, and data scientists don’t like what I say, my belief is that every chart is a manipulation. Every chart is a series of decisions about what to show and what not to show. And so we have to think about how we make those choices responsibly and ethically without bending the truth too much.

Nancy: Is there a right time to show all the data versus only showing the data that supports your case?

Scott: I think the earlier on in the process and the less you know about what you’re trying to show, the rawer the data could be and should be. It can always be raw if you’re working with experienced people, and I talk a lot in the book about context setting. If you know the people you’re working with understand the data, and understand then what you’re trying to accomplish, and can work in complex visuals, go ahead and keep it raw.

Nancy: On the other hand, you talk about simplicity being really difficult. Why is simplicity so difficult?

Scott: It’s a really hard thing to show somebody a simple chart for a couple of reasons. One is sometimes we’re not sure what we’re trying to say or the best way to say it, so we put everything in there and hope that it sort of emerges. We don’t really hone our message. But the other is complexity looks like busyness. It looks like work. And sort of looks like saying, “Look at all the work I’m doing. Look at the data I have. All the data I’ve collected.” Simplicity has this effect on the audience where they’ll see something and understand it so quickly and so fully that it almost feels like they never didn’t know it. So, that’s a hard thing sometimes to present because people worry people’ll think, “Well, he didn’t do any work. That’s so simple and clear.” And I just think that that’s a misperception. We have to embrace simplicity.

Nancy: I loved your thinking around non-persuasive and persuasive charts. Wouldn’t all information fall into the persuasive category? Why prepare information if you’re not trying to convince someone of something?

Scott: You’re never not persuading. It’s just how we operate as humans, right? We’re constantly being persuaded, and we’re constantly doubtful. [Even] showing a perfectly objective—and there’s no such thing as a perfectly objective chart—but showing a chart that’s aggressively objective, that just says, “Here’s the data we’ve found. Here it is sterile,” is a persuasion technique. You’re trying to persuade something that you’re objective and credible.

Nancy: You have a whole section on how data tells stories. I was just curious about how you define stories and how does data tell stories?

Scott: In a lot more cases than people think there are stories in data. If we break down stories to their fundamental atomic elements, every story is essentially setup, conflict, resolution. Here’s a reality. Here’s something that happens to that reality. And then here’s some new reality, some resolution.

But a lot of charts have this in it. Especially when you think of time-based charts. You have some trend, something happens to the trend, and then something happens after. And you can break charts down this way. And I think that a lot of times when you get those complicated charts, one of the things that you can think of with complicated charts is that is there actually a simpler story I can create by creating a set up, creating the conflict by showing a new chart, and then creating the resolution. We love narrative. There’s all kinds of research that shows narrative is so powerful of a tool. And I took two simple examples in [the book]. One was a simple line chart of the cost of peanut butter. You couldn’t think of something simpler, but you can actually turn one simple chart about the cost of peanut butter into five or six charts that tell a story about the cost of peanut butter.

Nancy: The other thing is the book ends with a visual critique section. It asks you a bunch of questions and lets you go through your own exercise resketching the charts, which I thought was really cool.

Scott: I’m pretty passionate about that, too. This really resonates when I talk to people. A lot of people know or say that they want to get better at data viz, or visualization, visual communication, but they’re sort of afraid because the visualization community can be pretty intimidating and pretty judgy.

Nancy: Judgy? I agree, man.

Scott: If you’re on Twitter and you see some of the tweets about so-called bad charts, it’s really destructive criticism. And I really wanted to bring it back to the classic design. I’m always disappointed in some of the stuff that goes on online with other people bad mouthing other people’s charts without even understanding those people’s context.

Scott Berinato is the author of Good Charts. Follow him at @scottberinato.

Nancy Duarte is the CEO of Duarte, Inc. (contact@duarte.com) Nancy is the co-author of the recently released Illuminate: Ignite Change Through Speeches, Stories, Ceremonies, and Symbols and the author of Resonate, Slide:ology, and the HBR Guide to Persuasive Presentations. Follow Nancy on Twitter at @nancyduarte, and purchase Illuminate here!

Graphic credit: Scott Berinato, Harvard Business Review