Introduction
How Do I Start?
Hello.
You may have picked up this workbook after reading Good Charts: The HBR Guide to Making Smarter, More Persuasive Data Visualizations—my previous book, which offered a framework for understanding what makes good charts and laid out a process for creating them yourself. Or you may have picked it up in a shop just because dataviz intrigues you. You want to make good charts—or at least you think you should be able to. Maybe you got here through an online search. Or a colleague, a friend, or your boss handed you the book because he or she knows you like to think visually. In any case, you’re here. And you probably have the same question most people ask once they decide that data visualization is something they want to learn about: How do I start?
When I speak or lead a workshop on data visualization, audiences are easily inspired by the transformations I show, and they understand the core argument of Good Charts—that what makes a good chart is not how pretty it is or how well it follows some set of chart-making rules but how effectively it conveys ideas by adapting to the context in which it will be used. But inspiration can be short-lived. Many feel overwhelmed by the idea of doing it themselves. So they ask me, How do I start?
Start here.
An analogy: For years I wanted to learn to play the guitar. I was inspired when I watched a friend play or heard a song with a deft guitar line. But I never picked one up, because I felt that same dread: I didn’t know where or how to start. Finally, inspired by my daughter—who took up the guitar (and got good fast)—I decided to just start. With the help of a workbook, I learned notes, and then notes became chords. Eventually I added strumming patterns. Before long I could play a few simple songs, such as Bob Marley’s “Three Little Birds” and Etta James’s “I’d Rather Go Blind.” I continued to build my skills and repertoire through practice, and although I’ll never be a master of the instrument, I can work my way around it. In truth, it didn’t take as long as I thought it would, and it wasn’t as daunting to learn as I feared it would be. I just needed to start.
Good Charts Workbook provides the ideas and exercises to help you practice dataviz. They are its notes, chords, and strumming patterns—the foundational concepts and approaches that will soon have you playing simple songs. The workbook will help you understand why certain approaches to chart making work or don’t work and prompt you to think through challenges yourself. It will allow you to test your ideas, and it provides a discussion about each challenge to help shape your thinking and build your dataviz literacy. It sets a foundation that will make the process of creating good charts as automatic for you as it now is for me to switch from a G chord to a D chord.
Let’s keep this lo-fi. Most of the work that goes into making good charts does not happen digitally. Charts I create tend to be about 90% complete before I start digital manipulation. To get the most out of this workbook, you need:
Blank paper. Extra paper will be helpful if you sketch the way I do—fast, messily, and over large areas. I don’t like to feel constrained when I’m sketching, so spreading paper out over a table helps. Extra paper will also allow you to reuse challenges with others or to go back to them with fresh eyes after some time.
Colored pencils. I recommend having only a few of these available while you’re sketching—say, a black one, a gray one, and two colors. (I use orange and blue quite often, but the choice doesn’t matter.) It helps to make them contrasting colors so that you have the basic tools to show both complementary variables that can be different saturations of the same color and contrasting variables that shouldn’t look as if they’re part of the same group. I find that when a chart has too many colors, I focus more on refining its color scheme than on the expansive process of fast, idea-generating sketching. Once I get to prototyping, though, and I’m trying to create a viable, realistic, neat sketch of the chart, I like to add colors. With this workbook, you’ll be both sketching and prototyping, so a set of about 10 colored pencils will serve you well.
Energy. Attacking these challenges when you’re tired or not in the mood will be a slog. Sometimes my best ideas come after I put the work aside for a while and come back to it in a better frame of mind. Solutions that seemed elusive suddenly appear. Anyone who does crossword puzzles will recognize this phenomenon. The answer to a clue that irked you is suddenly obvious after you put it aside for a bit. It’s the same with dataviz.
How is the workbook organized?
Two core sections make up the book.
Part 1. Build Skills
Each chapter in this part includes:
- •A brief introduction to a dataviz skill, including six guiding principles
- •A warm-up, including several small challenges to reinforce the guiding principles
- •Three core challenges, each incorporating larger-scale tasks that address several or all the guiding principles
The challenges in Part 1 are organized according to the skills they’re meant to develop. Their scope is limited in that they don’t ask you to create something from nothing. In many cases the context (or multiple contexts) will be provided for you. The challenges are designed to focus your efforts on one skill at a time. You can flip to any challenge in the book—be it a warm-up or a core challenge—and try it, just as you could flip through a crossword book and pick any puzzle. Before you take on a challenge, though, it’s helpful to read the chapter introduction and think about the guiding principles. Highlight key ideas from them. Everything flows from those principles, so it will be hard to get into the right mindset without having thought about them.
And although you don’t have to tackle the challenges in order, the book does follow a loose logical progression, from more-fundamental skills (color, clarity) to more-complex ones (persuasion, conceptual charts). It’s not a hard-and-fast pedagogy, but you may find it helpful to start at the beginning before jumping around.
Immediately following each warm-up section and each challenge, you’ll find a discussion about it that includes my effort at solving it. I’ve deliberately avoided calling this an answer key. That’s because I don’t presume to have the right answer to any of these challenges. The charts you come up with could be completely different from mine and just as, or more, effective. In some cases I admit to being unsatisfied with my final approach or talk about the trade-offs I made to arrive at it. That’s OK and entirely typical. It’s rare that you don’t have to make a trade-off to create a good chart. The discussions are not meant to tell you the answer; they’re meant to expose my thinking to help guide yours.
Part 2: Make Good Charts
This part provides two large-scale challenges that require multiple skills from the previous section. They enlist the talk-sketch-prototype framework from Good Charts and are bigger and more open-ended than the previous challenges. I recommend that you save them until you’ve tried some of the skills-building challenges.
Just like the Build Skills section, discussions including my attempt at tackling them follow these big challenges.
In addition to these main sections, you will also find appendixes to help steer your efforts. Good Charts Workbook uses many chart types and reveals how the visual words and phrases you use to describe your data (“spread out,” “a portion of,” “distributed”) may suggest a chart type for your given situation. To that end it includes some reference materials that show chart types, use cases for them, and some of the keywords associated with them. (These materials also appear in the original Good Charts.) They’re excellent tools to have handy when you’re in the process of talking and sketching. Wear out the back of this book looking at chart types and use cases and making notes about them.
How should I use the workbook?
First, I urge you to avoid short-circuiting the challenges—that is, don’t read a challenge and then immediately flip to the discussion to see how I approached it. The workbook is meant above all to help you think for yourself about data visualization. Don’t bias your approach by first looking at someone else’s. No peeking! Hell, if it helps, tear out the discussions and put them elsewhere.
The skills-building challenges are focused and contained, but expand on them if you like. If you’re working on a clarity exercise but see an opportunity to build some color skills, go for it. Want to create a new context for a challenge and then create a chart that reflects it? Go. In my discussions you’ll see that ideas from multiple chapters pop up in any given challenge, because none of these skills can be completely isolated. Sometimes ideas about color find their way into a challenge about clarity. A persuasion challenge may require some clever choice of chart type. Use whatever you learn wherever you can.
Many of the discussions will include well-designed “final” charts, but you’re not expected to create final products in this space. In most cases sketches and paper prototypes, or very neat sketches that approximate a final chart, are as far as you’ll want to take your work. Get to a good idea and a good approach, and you’ll have developed the material to create that final chart. As noted, most of the work that goes into making a good chart happens before you use digital tools to create the final product. Of course, if you want to work on your production and design skills as well, go for it.
On the charts and data in this book
Some of the charts in this book are obviously real. Others are based on real charts but have been changed substantially, whether in subject, data values, colors and labels, or any number of other elements. I’ve done that for several reasons—sometimes to protect proprietary data, other times to make the challenge more difficult or to change the context being addressed.
Several charts from Harvard Business Review and HBR.org are included here with the kind permission of Harvard Business Publishing. In some cases they have been reverse engineered to be poorer versions of what was ultimately published. That’s for learning purposes only. The published versions are good charts; these rigged versions reflect neither the authors’ nor HBR’s intent.
Finally, some of the charts included here are simply bad: they suffer some fatal execution flaws or they’re just a bit of a mess. Presenting something suboptimal gives you an opportunity to learn from and improve it. But it’s important to note that while some of these charts are not ideal, they are realistic, in that they employ common approaches and techniques that I’ve seen in the world, online, and in my work helping others with their dataviz.
On tools
Next to How do I start? the most common question I hear is What tools should I use?
The answer is unsatisfyingly complex: No one tool works well enough to be the dataviz tool; many dozens exist, and more are coming online all the time. They all do some things well, and none of them does everything well. The more complete, more powerful tools—usually meant for data scientists—have a much steeper learning curve than those available free or for a small fee online.
I have about six to eight tools I use regularly, and I reevaluate every so often as new ones come online. I’m hopeful that soon we’ll have good tools for non–data scientists that make the answer to this question much simpler. I’ve seen some tools in development that look extremely promising but are far off.
Search online once you know what kind of chart you’re trying to make. Experiment with several tools and learn what you’re comfortable with. Bookmark the ones you like. And remember, nothing beats a pencil and paper. You can get most of the way to a good chart with some talking and sketching.
I also advocate having another tool at your disposal: friends. If you know good data wranglers and good designers, or if your organization employs them, use them. I maintain a kitchen cabinet of friends and colleagues I rely on to help me with advanced data and design challenges. Dataviz is complex; it should be a team sport. More and more organizations are setting up teams to take on important visualization challenges. Put together a subject-matter expert, a data analyst, and a designer, and you’ll up your dataviz game significantly.
One more thing you might find helpful is my workflow for creating (and re-creating) the charts in this book. Although I worked mostly on my own, I did lean on some people to help me, and you can imagine where in this process I’d bring them in.
- 1. I used an app called Sketches on an iPad Pro to take notes, sketch, and prototype. This was my “paper and pencil.”
- 2. The data for this project was mostly stored in Excel or CSV (comma-separated values) files, where I created typical Excel visualizations just to have some initial view of the data.
- 3. I exported that data into an online tool called Plot.ly to re-create the initial Excel chart and manipulate it there. I also exported images from Plot.ly that I could import back into Sketches for marking up and discussing.
- 4. I used that same Plot.ly work space to create all the digital prototypes that came out of the sketch session. Those prototypes are generally not pictured here, because many are remarkably similar and wouldn’t show enough progression to warrant the space they’d take up in the book. It’s not uncommon for me to produce 10 to 12 very similar prototypes for a chart as I refine it.
- 5. When my digital prototypes felt close to complete, I exported SVG files from Plot.ly and imported them into Adobe Illustrator, where I have templates set up for my typography, colors, and other design standards. It’s here that I polish my designs.
Finally, I’d love to see what you create from these challenges and from using the talk-sketch-prototype method. To that end, I’ve created an email address ([email protected]) where you can submit your charts—the befores, the afters, or both. Part of being a good chart maker is being a great chart user. Seeing others’ work can be inspirational; sharing is a core ethic of the dataviz community. Even so, sometimes sharing leads to unwanted critique. The dataviz community can be oppressively judgy. That’s not the goal here. I will not publicly critique anything you submit to me without your permission, ever.
OK, you’re ready. Wear this book out. Scribble notes all over it. Highlight things. Mark it up. Copy stuff. Take notes. Fill it with your ideas, your favorite approaches, color schemes you think are effective, and chart types you’re particularly fond of. Critique my discussions. In short, use the workbook strategically, but really use it. You can always come back to it for inspiration or to look something up. I hope that when you’re finished with it, this workbook is uniquely yours.