Understanding The Big Picture
A conversation with data visualization expert and author, Steve Wexler
Steve Wexler is a data visualization consultant, author, and head of Data Revelations. He co-authored The Big Book of Dashboards: Visualizing Your Data Using Real-World Business Scenarios, and recently wrote the new book, The Big Picture: How to Use Data Visualization to Make Better Decisions—Faster.
You began your career as a musician. You compose as well as play, right?
Poorly. My focus was arranging and orchestration. You tell somebody you’re a data visualization consultant or data visualization specialist. It’s like, what’s data visualization? What’s arranging and orchestrating? It’s taking a composition and repurposing it and re-rendering it, into something that is bigger, different, and hopefully better than what it was originally. So yeah, I’ve composed. I would consider myself mediocre at best. A reasonably good arranger and orchestrator, and a competent bass player.
“What’s arranging and orchestrating? It’s taking a composition and repurposing it and re-rendering it, into something that is bigger, different, and hopefully better than what it was originally.”
Hear about upcoming conversations:
You talk about creating a visualization for a particular audience. Does that play into when you're arranging music?
All these things involve problem solving, and whatever the way these things tickle my brain in doing a data visualization or a dashboard or crafting a presentation or a big band arrangement or a symphonic suite or something like that, there’s problem solving that's involved.
And people might say, “Well, wait a minute, with music you’re trying to convey emotion and feeling.” And I say, “Well, you probably are with the data visualization as well.” But figuring out, “Well, how is this going to work and how is it going to be interpreted by the audience? Gee, I can’t just start with everything blaring.” That’s like “Let me just show 15 visualizations with each one having 22 different colors in it.” Well, where are people going to look and what are they going to focus on? It’s the same thing with music. Maybe you want to start softly. “I want you to hear just this little thing. It’s going to capture your attention.” So, the problem solving aspects of it, I find remarkably similar.
Color on the left highlights something interesting about the data; color on the right obscures any insights about the data. From The Big Picture: How to use data visualization to make better decisions—faster.
I know you started your career with music and then discovered data visualization, specifically with survey data. Is that correct?
That was where I got my start. I ran a software company, then founded a startup but couldn’t keep funding it, so I found myself in a position where I had to get a job for the first time in my life, and ended up at a wonderful place called the E-Learning Guild, (now called the Learning Guild). It’s an organization that’s passionate about using technology to help people learn and perform better. They had thousands of members and were going to tap into their collective expertise and survey them, visualizing the results. My big epiphany, and I think this got me the job, was that instead of static reports, we could have live interactive dashboards.
If people don’t want to see all responses, and they just want to see responses for their industry or for a non-profit versus a commercial entity, they could do that. And that’s what got me into data visualization, and I was pretty damn bad at it.
“I think the Minard chart has caused an awful lot of damage. It is a brilliant piece of work, but this is not something that you should be trying to emulate in business visualization.”
It was kind of a trial by fire then, needing to gain that skill to make it happen?
There were a bunch of things all competing, both getting good at the tool we were using and getting good at data visualization. It’s when I discovered Stephen Few’s Now You See It and Show Me the Numbers. Up until that point, I had read some Edward Tufte books and thought, “Gee, you’re showing me these charts by Charles Minard and Florence Nightingale and yeah, they’re wonderful, but how is this going to help me visualize survey data?” And the answer is, it’s not. I have a problem, as you know, with the Minard chart as I discuss in chapter eight of The Big Picture. I think that chart has caused an awful lot of damage. It is a brilliant piece of work, but this is not something that you should be trying to emulate in your business visualizations.
Charles Minard’s 1869 “Figurative Map” of Napeoleon’s campaign to conquer Russia in 1812-1813.
Interestingly, Minard was a civil engineer, and he created many other charts and so many of them are way more applicable to business today than this one-off, this brilliant, but strange concoction. In any case, I interpreted Stephen Few’s writing as, “I’m trying to help you as a businessperson create business visualizations that are going to lead to insight.” Ah, that’s what I was looking for!
Do you feel like you learned the technical part first and then the design skills emerged over time?
They were both happening at the same time. I was getting better at understanding how other people understand data pictures as I was learning the technical part. “Oh, you know what? If I could show it this way, I think people will understand it better and faster. Well, how do I make that thing?” If I want to compare this group with that group across 14 categories, this is an easy thing to do in Excel. It’s not so easy to do in Tableau, but then someone says, “You know, there are other ways to do this. You could do a bar in bar chart. You could do a gap chart. If you’re showing change over time, you could use a slopegraph or a comet chart.”
A Comet Chart is an alternative to a paired bar chart. From The Big Picture: How to use data visualization to make better decisions—faster.
Now I might ask “How does Pew Research show this? How does The Economist show this? How do other organizations, when they’re saddled with the same problem, elect to do it?” then say, “That’s a really good way to do it. Now how do I do it in my tool of choice, whether that be Power BI, Qlik, MicroStrategy, Excel, or Tableau?”
At what point in this journey did The Big Book of Dashboards come about?
It came about towards the end of 2014. A client was working with survey data, and they asked, “We’re doing this thing in Excel, and it takes three people working full-time for four weeks to produce this monthly report. If we did this in Tableau, could we automate it and do it faster?” And my answer to them was, “Yes, but there are better ways to show this than what you have here, which is just this kind of trellised red, yellow, green stuff all over the place. There’s a way that you can glean better insights faster. So if you’re open to a new tool, why settle for making bad stuff, faster? Why not make something that is a lot better?”
The catalyst for The Big Book of Dashboards. This is NOT a good way to present the data.
They said, “Really? We think this is really great—nine different regions trellised with lots of green, yellow and red.” I said, “Let me take a stab at this,” and I built some prototypes and I wrote a whitepaper showing, Here’s how you’re currently doing it. Here are the shortcomings with it, what you’re missing, and here’s my proposed alternative and why it works better.” They loved it. But…
Their client didn’t. They said, “No, do it the way we’ve always done it.”
This is a battle we lost. But this, “Gee, if you have this situation and this data and this predicament, here’s a good way to show it and why,” that’s the foundation of The Big Book of Dashboards. There are all these different scenarios, and the idea behind the book is that you look for something that is close to what you’re trying to do, and we’ll show you, “Here’s a solution, here’s why it works, here’s some alternatives, and maybe you can take this and apply it to your own work.”
How did you get hooked up with your co-authors and where did the idea of writing a book come from?
Well, I was hoping that these other people would do all the work and I would get the credit, and wondered, who could I hoodwink into doing the heavy lifting while I get all the applause? No—really, for whatever reason I didn’t feel that this was something that I should do on my own, and it was just a brilliant stroke of insight because it would not have been nearly as good a book had it just been my own.
The first person I asked to do it was a woman named Kelly Martin. She was a Tableau empress, and she had really great design chops and an uncommon analytic mind. She hated when people would think of her as just, “Oh, she’s the really great designer,” when there was much more to what she did. I had also come to like the work and approach of Jeffrey Shaffer, and convinced Kelly, “What do you think about having a third author on this, Jeff Shaffer?” By the way, he’s a fellow musician and brilliant trumpet player, MBA, and CFO at Unifund. A very accomplished cat. Kelly had to drop out of working on the book. Too many other conflicts and difficulties with her health. She died just about two years ago. Really brilliant, brilliant work. She showed people who were using Tableau at the time, “Oh, you’re all working at this level. I’m going to bring you up here,” and showed a stunning design approach, humor and humanity, and just lifted us all to new heights in terms of analytic integrity and superior design.”
“I kept saying, ‘Do you want to go down as the data visualization equivalent of the guy who turned down The Beatles?”
In any case, she was unavailable to do it. Jeff was still in, and Jeff and I agreed, it’d really be good to have a third voice in this thing, and we realized Andy Cotgreave would be great. We had to convince Andy to do it. I kept saying, “Do you want to go down as the data visualization equivalent of the guy who turned down The Beatles?” We were able to convince Andy, and he was an incredible collaborator. So I chose two strong people, and we had dissenting voices, and that was great. People ask, “How do three people write a book? Didn’t you get into fights? Didn’t you have arguments?” I would say we had debates and discussions, but it wasn’t like a band that was going to break up because we were having artistic differences. It wasn’t like one of us saying, “Oh man, I want to play jazz. I don’t want to do this pop stuff any longer.”
We were all seeking clarity, or what I’ve now encapsulated as, “we are here to provide the greatest degree of understanding with the least amount of effort.” Jeff started with, “Who’s your audience? What’s the message?” And then for me, “for the largest number of people in that audience, provide the greatest degree of understanding with the least amount of effort.” It was a very good experience working with the two of them.
When did you start with doing workshops?
If it were up to me, I’d probably still be thinking, “I don’t know if I’m ready yet.” Within a month or two of The Big Book of Dashboards coming out, Jeff said, “We can do this,” and I did the first couple with him. Realize he has two full-time gigs, and he can’t be out there giving workshops that often. This is mostly what I do, workshops and classes and presentations related to the books, and he helped really jumpstart this thing and held my hand on it. Within three or four months, I was doing them on my own. Every now and then we’ll do one together.
“I’ll tell you what I think is the number one infraction in data visualization… It’s the misuse of color.”
What's been the most difficult concept for your workshop attendees to grasp?
I’ll tell you what I think is the number one infraction in data visualization. It’s the misuse of color. Not understanding how to use it, where to use it and how to be purposeful with it. We make the case in the book that there are really only five ways to use color in data visualization, categorical, sequential, diverging, highlight and alert, and alert is really just a special case of highlighting. And highlighting in turn is just really a special case of categorical color where everything is the same except this one thing. And people don’t get that. They think, “Oh, wow, I’ve got eight bars? I’m going to use eight different colors.” Recognition of how to use color intelligently and sparingly can be a game changer for drawing people’s attention. I think that’s the biggest eye opener.
The five ways to use color in data visualization. From The Big Picture: How to use data visualization to make better decisions—faster.
You mentioned Stephen Few, and one of the things I've noticed is that he has very specific definitions for what a dashboard is and isn’t. What do you think?
Stephen beat us up for not having a more rigid definition. We define it as “A visual display of data used to monitor conditions and/or facilitate understanding.” He thought if we had just said ‘or’ we would have been fine, but we said ‘and/or’ and now it loosens things up.
We went through a whole hullabaloo of static vs. interactive, multi-screen vs. single screen, exploratory vs. explanatory and down this rabbit hole. There’s a disciple of Stephen’s named Nick Desbarats who runs Practical Reporting based in Canada. He has taken over doing the workshops that Stephen had done. He did a presentation at the Tableau Conference in 2019 in called “13 different types of displays that are unfortunately all called dashboards.” He was outlining this taxonomy of the different types and what they fall into, and I really didn’t buy into this, but …, he’s not saying that the audience must buy into it, but that we, as practitioners, need to help manage people’s expectations. If someone is thinking that they’re going to get a static report with highlights, and instead, they get an interactive dashboard where they’ve got to discover stuff for themselves, they’re going to be disappointed.
Likewise, if they’re expecting to be able to drill down and explore things, and instead they get a static report, they’re going to be disappointed. I think Nick is spot on with this.
“I think it’s fine for a dashboard to be boring. But the presentation you can create on some finding from the dashboard? That should be riveting.”
I don’t see eye to eye with Nick on everything, but I do on a lot of things. His four-threshold concept really deserves attention as it makes it easy to see what is in crisis, what needs attention, what is doing well, and what is extraordinary.
Desbarats’ “four-threshold” approach
In any case, we intentionally chose a somewhat open-ended definition of a dashboard. The dashboard that you use to help you understand the data may never be shown publicly. You may say, “I found something cool. Now I’m going to create this badass presentation with a visualization that is totally divorced from the dashboard.” That’s okay.
The other thing that I’ve said that may be somewhat controversial is, I think it’s fine for a dashboard to be boring. But the presentation you can create on some finding from the dashboard? That should be riveting.
Can you talk about the importance of collaboration in data visualization?
I didn’t buy into the idea dashboard by committee until working on The Big Book of Dashboards and getting pushback when stuff that looked clear to me didn’t look clear to my fellow authors. It also helped for me to watch a colleague run a clinic to increase data visualization skills. What’s a good way to show this stuff? It was high stakes in that they were embedding dashboards as part of what they were selling, and I thought “This is going to be a disaster. You’re going to get people who say, ‘No, you should do it this way.’ and ‘Can’t we make it a pie chart and make it 3D because that looks cool?’”
It was an opportunity for me to bring them up to speed with, “What are we trying to do here and which of these things answer the questions faster?” Nick Debarates talked about the same thing, which is, don’t come in as the expert, saying, “Well, I know what I’m talking about you’re an idiot.” It’s, “Well, what is the question we’re trying to answer?” We’ve got three possibilities, and which of these is the easiest to understand? And at the same time, you’re starting to see stuff through the stakeholders’ eyes and understand their business problems. They start to learn foundational elements of data visualization, and you start to learn some of the challenges the stakeholders have, and together we end up making something a lot better than what we would have come up with separately.
“Collaboration isn’t me saying “red”, you saying “blue” and us agreeing on purple. That’s compromise. Collaboration is when we make something together that is better than what either of us would have made separately.”
There’s an aphorism that I have in the book, and I woke up one morning convinced that Stephen Sondheim had said this. It’s, “Collaboration isn’t me saying “red”, you saying “blue” and both of us agreeing on purple. That’s compromise. Collaboration is when we make something together that is better than what either of us would have made separately.” I spent forever trying to find this quote, and then trying to figure out, if it’s not Sondheim, who said it? Apparently, it’s me!
There’s plenty of times that I think it’s better off if I do something on my own, but anything that’s high stakes, I’ll show it to somebody and ask, “Is it as clear to you as it is to me?” And I will often get some good ideas about how to improve it. There are a lot of people who did deep reading on The Big Picture, and I purposely sent some chapters to people I knew wouldn’t necessarily like what I was trying to say. I wanted to know, “Where are there flaws in this and why?”
Are you happy with the reception of The Big Picture? Is it starting the conversations you were hoping for?
It is starting the conversations I want to have. It’s not a book for data visualization practitioners, although I’m discovering a lot of them are finding it helpful. When I did a workshop two weeks ago, the person who commissioned me said, “I want this to be a pre-read for everyone who’s taking the workshop, and I want them to come in with this understanding.”
“Everyone needs to understand this, appreciate it, use it day-to-day, and know when they’re being bamboozled by a misleading chart, and so forth.”
I’m trying to get this to be something that Human Resources and Learning Development realize everyone in an organization needs to understand. Maybe not everybody is going to become a writer of books, but they all need to be readers of books. Everyone needs to understand this “language”, appreciate it, use it day-to-day, and know when they’re being bamboozled by a misleading chart, and so forth.
The uphill challenge was articulated by Cole (Nussbaumer Knaflic) when she was reading an early draft of the book. She said, “You know, Steve, your challenge is going to be that the people who really need to read this book don’t know they really need to read this book.” So, I am trying to get the practitioners to evangelize the book as well. If your organization doesn’t value data visualization, they’re not going to value the data visualization practitioner. You, the practitioner, have the ability to transform the organization, but only if people recognize what you bring to the table.
A big catalyst for writing the book came from the frustration of workshop attendees. They understood the value of data visualization, but so many of their stakeholders didn’t. I think the most crushing thing you can do to someone who has worked hard to make a brilliant dashboard is say, “Hey, could you add a button that’ll allow me to export the data as a …?”
I can think of one case where the desire to download the data is a good thing: the stakeholders are so inspired that they want to explore the data on their own. But so often the request is because the stakeholder doesn’t YET see the value in the data visualization or dashboard. But if the consumers can learn the language of charts, and if the dashboard creators understand the needs of their stakeholders, truly transformative things can happen.
To learn more about Steve Wexler, visit Data Revelations