Transcript

Speaker: Welcome back to the Policy Viz Podcast. I'm your host, John Schwabisch. Hope you're well. Hope you are enjoying season 12 of the podcast thus far. On this week's episode of the show, I welcome Justin Evans, author of The Little Book of Data.

Speaker: Actually, not that little, you know, it's about 250 pages or so. ah really enjoyed this book. This is a nice read. I have to tell you, subtitle to the book is understanding the powerful analytics that fuel AI make or break careers and could just end up saving the world.

Speaker: I'm not sure about saving the world, but it is an enjoyable book. Justin does talk about his experience working in a variety of different organizations and sectors. He does talk about what it means to be a data person. And we talk in this episode about...

Speaker: whether anyone can be a data person or whether we need special skills or special insight or special training. ah We also talk about some of the most effective ways people and organizations can build data and data visualization literacy within with their staffs, which is something that Justin has spent a lot of time working on.

Speaker: We also talk about dashboards. It's something that's sort of been rumbling around in my brain over the last few months is whether or to what extent dashboards are useful. Public dashboards are useful. When we post these data dashboards out to the world on websites, do people actually use them? Do people actually download the data and use them? Do they make the graphs and and use them? I'm not really sure.

Speaker: I've been thinking about this for a little bit. There is an interesting paper out of Tableau Research ah showing that at least in there, I think like two dozen or maybe 30 people that they interviewed that most people didn't actually use the dashboard, that they explored it for a second and then they downloaded the data and did their own analysis.

Speaker: So Justin and I talk a little bit about dashboards in this conversation because it's something I've been thinking about a lot. um We also talk about some of the challenges that are happening at the federal level when it comes to data.

Speaker: And we talk about being able to trust the producers, collectors, and communicators when it comes to federal or government data sort of more generally. um Again, I really enjoyed this book. It was a nice little read. i got to read it over the summer, which of course is always nice when you are reading during the nice weather.

Speaker: um So let's ah turn it over. Here's my interview with Justin Evans, author of the new book, The Little Book of Data. Hey, Justin, good to meet you. Thank you, John. Great to be here with you.

Speaker: ah Thanks so much for for coming on Little book of data. I've got the fun galley copy, so there's no index. I had to put notes in the book, which I hate to do. I had put notes in the book to make sure I knew where to go back to.

Speaker: um You have to have that classic undergraduate ballpoint pen, perception versus reality note. Yeah. Yeah. Yeah. And try not to highlight every sentence because but um so I appreciate you coming on the show.

Speaker: So I thought we'd start sort of like the basics, like, you know, who are you, what's your background? And then, and then maybe just get right into the book. Like what is the, you know, what is the overarching sentiment of the book and who do you think will benefit most from reading it?

Speaker: Well, again, thanks for having me. I'm Justin Evans. I wrote the little book of data. I wrote it about 20 years into my data career, which started at the Nielsen company, the big TV research company.

Speaker: And at the to leap forward a lot, I was about 20 years into my career and working at Comcast, a big cable company in one of their advertising divisions.

Speaker: And what I was observing is that a lot of mid-career people who are really rock stars kind of in the old traditional TV world were bailing out of their careers or being bailed out.

Speaker: And what I was noticing is that the people who embraced data and that next generation of marketing and advertising that was based on data and machine learning were persisting in their careers and succeeding. And the people who couldn't embrace it for whatever reason were parachuting out.

Speaker: And actually found this kind of distressing because the people that I was seeing leading the business were great people and great professionals and great people leaders. And it made me think, what's the what's the matter with data that people can't get themselves geared up to learn it?

Speaker: And i started to sort of investigate that and think about the what I call the four layers of data denial, which is that you – you think it's too hard or you don't have to learn it or it's too complicated or you're just intimidated by it.

Speaker: And I thought to myself, well, what, what if I could write a book that was almost written for that person, the person who is a leader or a general business person, but who doesn't understand data, but has an inkling that, you know, this, this data thing could be powerful and could mean something in my career.

Speaker: So, so I ended up, kind of writing the 20-ish core ideas that I felt were really true of data that would want to tell that person or tell my younger self.

Speaker: And then try to collect stories about those ideas. And and full full transparency for your listeners, I wrote a very bad first hack at this that was I'm just going to tell this data idea in the simplest possible language.

Speaker: And my friend read it and said, this is just, you know, not that it was unreadable, but it's not fun. Make it fun, tell stories around it. So what I did was I then started the process of trying to find people whose careers, whose work illustrated these 20 ideas. And that went much better and ended up being a ah complete joy meet these data professionals who I regarded as sort of heroes in a way.

Speaker: and hear their stories and turn those stories into the heart of the book. So you're learning the lessons of data, right? You're learning how to think like a data person, but you're not being burdened by the technology and code. You're actually just learning it by seeing how other professional data people think about data problems.

Speaker: Right, because it's the thinking about data problems that's the first step. That's right. Like you get into code later, but but the the the human part of it, i mean, we're not gonna talk about AI. i can't I can't talk about AI anymore, but like the AI part of it is like a whole other thing, but like the human part is the critical thinking about data.

Speaker: That's right. And I would add the the problem solving part of data is really the essential part of it. I think even for people who do it every day,

Speaker: and do it for a living. i mean, i can't code my way out of a paper bag and I'm really not even that strong with math, but I've been in the data business a long time. i am a data person. I think like a data person and there's almost no business problem or a life problem that you can't put on me that I can't turn into a data opportunity.

Speaker: And it's fun. And if you have to, once you get some of the kind of core principles in your head, then it starts to flow. And in my experience, and I'm not diminishing the importance of people who are quantitative and are coders and are math people, but generally it's the vision about what you can do.

Speaker: and why you need to do it, that drives a process. And if you have a vision about what and why you need to do, then you can find the technical people to make it so. Right.

Speaker: So I wanted to ask you about the data people, because you have a whole chapter in the book entitled Data People and Why I Love Them, um which was one of my ah favorite chapters. I'll also say, just going back to your part about stories, like that's why that's why I was able to read the book like by the pool this summer, because it was just kind of an enjoyable read. It wasn't like...

Speaker: It does get in the weeds, but, but you're seeing how people have had to address these challenges with data. So I'm all in on the storytelling to sort of get that message across. But I wanted to ask about, about data people.

Speaker: And you've already alluded to this a little bit but like, do you think people are intrinsically data people? Do you think it's a skill that needs a four year degree?

Speaker: Like, like what makes a data person, a data person? What I tried to do in writing that chapter was think about all the people I had hired over the years, which is probably in the hundreds and try to think about what made, what was similar between those people.

Speaker: And especially when I was leading teams that were going through a period of transition where we had to let go a certain kind of person and hire a different kind of person.

Speaker: And upon that reflection, I just thought that the the data people I had hired had a sense of what I call a, well, I'm not coining the phrase, a fiduciary duty. I'm probably applying it to the data world maybe for the first time.

Speaker: That people had a duty of faith to the client and a sense of mission. And they just, they really cared about getting the right answer for the client.

Speaker: And there was a sort of thrill in that responsibility and a thrill in the adventure of going into data and trying to come back with an answer.

Speaker: And the duty of faith part also meant that there was an ethical component as well. And when I think about the people that I exclude from the honorific of data people,

Speaker: I exclude people who just love the fact that they know more than somebody else. Especially in TV research, there's this class of person who could quote capture verse on Nielsen methodology, on the Nielsen TV ratings, but they couldn't tell you how to solve a business problem.

Speaker: And in a different end of the spectrum, heading west towards Silicon Valley, There are people who can tell you how to steal intellectual property and use data unethically and addict people against their will while using data in a clever way, but their lack of ethics excludes them from my honorable title of data person as well.

Speaker: yeah And actually had someone on my team Now, the other day, read that chapter and say, oh, that that chapter really describes me, especially the part you write about money.

Speaker: And I make the observation, I think I think I'm right about this, that I've never seen a data person rise to be CEO. You'll see financial people rise to be CEO.

Speaker: You'll see salespeople rise to be CEO, occasionally you're marketing a marketing or product person, but you never see a data person rise to be CEO. And I think that's because money is so final. You can't copy money and then try another whack at it. You know, you've already you've already gained or lost it.

Speaker: So data people definitely have, again, in my personal estimation, a certain dream equality that is, again, my eyes, laudable and even essential.

Speaker: They have to be able to close their eyes and and think about what's possible. But that sort of <unk> connection to reality can sometimes be a liability in other parts of their business life.

Speaker: Yeah. It's interesting the way you describe that. ah You didn't use a phrase like a programmer, a mathematician, ah statistician.

Speaker: It's almost like you take more of a humanities view of what it means to be someone working with data. I do. i think I think data people have to fall in love with the problem.

Speaker: and And once you do that, you'll do anything to solve it. You'll dream any dream and you'll work any hours. And working hard is part of the commitment. But indeed, the the hardcore quants and the hardcore coders, I i don't see that as essential to the to the the data personality, so to speak.

Speaker: Right. you You also spent a bunch of time kind of throughout the book talking about data literacy. I would throw sort of data visualization literacy and in there as well.

Speaker: um In your experience, what are some of the like more effective ways both people, but but sort of more broadly, I think organizations build that sort of data literacy skill ah throughout their their kind of their data workflow or data ecosystem?

Speaker: Yeah. so So you're asking what what makes, how how does one become data literate? Is that a good paraphrase of your question? Yeah. yeah The core ideas I really put in the book and tried to illustrate with some examples. And sometimes these things have to be learned by instinct and exposure. you know, one concept is the concept of identity, which is I can i can put a name on and an object, even an an abstract data object, and I can always come back and find that.

Speaker: And in the book, I tell the story of Scott Taylor, who is an employee at the Nielsen Company, who just kind of wandered into ah world where this trade magazine called Progressive Grocer, and B2B trade magazines, you got to love them. They're just, they just they're not cool.

Speaker: And he, there was, the the little label in the corner of Progressive Grocer had an identification number that was unique to the addressee.

Speaker: And that unique number actually meant that this person was the general manager of the Piggly Wiggly in Charleston, South Carolina.

Speaker: And it was associated with the store code. And what he found was that progressive grocer, this B2B trade magazine had the most comprehensive database of all stores, ah grocery stores and store locations in the United States.

Speaker: And it became this decoder ring for the manufacturers like Procter and Gamble and Unilever to know exactly where they were sending their cookies and their detergent.

Speaker: And they knew not just that it was a Piggly Wiggly, but it was store number 243 out of a thousand in the Southeast. And they knew that this guy who is the manager of the store was one person they had to reach, but they had 200 and some other, other general managers to reach.

Speaker: And therefore they knew how, what their sales penetration was. And they knew where they stood with that particular chain of retailers. And so this weird little label in the corner of progressive grocer magazine became a way to uncork millions of dollars of value for these manufacturers.

Speaker: And so if someone can wrap their head around a concept of identifiers like that, then they can really wrap their head around a core concept of data. And i think once you build up the idea of identification, the idea of matching from one database to another, the idea the idea of scoring different items in a database, you're you're kind of building up sort of the forehand, backhand, and serve of yeah of data, and then you can play the game.

Speaker: Right. Yeah. So let me get to the, to the other story that I love in the book again, also about Nielsen. i have in my notes, I think it was your friend, Freddie H. I don't know Freddie H, but, but that was the, that was the name of the, of the person building dashboards to sort of, to, to provide these insights on these data. And I'm curious.

Speaker: I guess on a couple of things, like on the data data visualization literacy, like what has your experience been sort of in how people improve their data visualization literacy? And also i'm I'm curious about your thoughts on dashboards, both internally and externally.

Speaker: um i'm I'm personally sort of going through this like thought experiment of like, or or experience really that internal dashboards have a lot of value because you and I work in the same company. We look at it we look at data in real time together and that's useful.

Speaker: But if I put it on a website, it's just another tool that you know just blows by most people and they're not actually going to use it. And so I'm curious in your experience about people using dashboards probably primarily internally um and then sort of building out people's data viz literacy in organizations.

Speaker: Well, the story, Freddie H, that story is actually a Samsung story. And this and the and the story was um the Samsung, we make money from selling advertising on streaming TV.

Speaker: And in 2020, when that story is set, the pandemic had just started. Everyone was now walked at home and everyone's, a billion people globally started streaming television overnight.

Speaker: And all of the marketers and advertisers and the ad agencies were calling us and saying, Samsung, you have a lot of data on people's TV viewing.

Speaker: What the hell is going on? how what Where are my customers watching TV? but Where can I reach them at ads now that the world is upside down? i really know how to reach anyone. Yeah. And the.

Speaker: It was one of those sort of emergency moments. I mean, it was much more fun than working in an yeah ER at the time, but it was an emergency moment in the ads business where we had to tell these clients immediately where to find their customers. and And because streaming TV libraries were so deep, people were spending lots more time with streaming and there's only so much linear TV you can take in that you want.

Speaker: And so we we ginned up a dashboard really quickly And it became a really great way to distribute a lot of data to a lot of people who needed it right away in in this aggregated form that told people answers to this question of where their audience is.

Speaker: yeah And if it was ah a moment where we created a lot of clarity for clients who were desperate and we created a lot of transparency or light in a dark room is a phrase I point in the book for clients who were afraid of or just unfamiliar with streaming TV behaviors and it made them feel safe as a place to advertise.

Speaker: So in that way, you know, at the time we were using the phrase democratizing data or simply sharing it at scale. I wouldn't say there was anything particularly strong about the visuals we created yeah at the time. But but in that case, it was it was a distribution mechanism that was what was powerful about it.

Speaker: But it also sounds like the reason it was successful is that people had fairly specific questions that they wanted to answer. That's right. And actually, but that's a really great hook into the essence of your question, which is why a why a dashboard and why a visualization?

Speaker: And the contrast you could make is to a dashboard, which is designed for i call, it a general user. In our case, it was salespeople who were generally accessing the dashboard, people who are ad salespeople who are not data people.

Speaker: And you can contrast that with a power user who's where where the interface is designed for someone to go in and really crank on data and it's ugly and it's hard to manipulate, but you can really go deep and ask it very refined questions in a dashboard.

Speaker: What's elegant and super fun actually about creating a good dashboard is that you are telling a story and you're guiding someone through a narrative that actually they may not have even known that they wanted the, at that time, the narrative that we created was if you,

Speaker: advertiser know who your audience is, I will tell you how they are watching streaming television. That was the question we were answering. And we kind of, it it was literally a ah flip book. We would say, okay, how many of your audience are there?

Speaker: It's 40 million. And flip the page. How many of them watch linear televisions? 20 million. How many watch streaming television? 10 million. How much time do they spend on linear television? I mean, I'm making it sound very boring, but the you you have to create a narrative. And I actually, actually I intend to do a ah workshop on this with my team. so thank you for reminding me, which is what what I force them to, when when we did that one and when we've done it since, what we do is we, I force them to use the English language and just ask the questions.

Speaker: How many of my audience are there? Where can I reach them? How much time are they spending? Can I even put ads in that environment? And we just, yeah we have these sort of rhetorical questions that are then answered by the data.

Speaker: And if those questions have a logical flow, then what's useful about that to the user is you've done the work to to to to tell the narrative. And it's efficient for the data team because if the data team only has to answer those questions.

Speaker: They're not creating some power user information. deep dive tool where you can ask it any question. They're only doing the analysis to answer those questions or setting up the data to be queried to answer those questions. So becomes efficient for everybody. And the beautiful thing is if you have ah team of people who are working front to answer those questions and doing that work and just shutting up, answering those questions and those questions only to create clarity, you have all these downstream clear clarifying effects.

Speaker: Right. if you do that upfront investment. Were these advertisers, I mean, you know, Netflix was hugely popular before the pandemic. Obviously, you know, all these streaming services exploded during the pandemic, but were they not deep into data prior to the pandemic? Was like, what, ah like what were, how were they making decisions without that sort of in depth analysis that you just explained that you and Freddie and others on the team sort of built out at the time.

Speaker: Well, in that and that moment, and actually we're still in it in the TV and advertising industry, there were only, there there was only Nielsen and similar data, which is based on smaller samples.

Speaker: And i don't want to get too deep into Samsung stuff, but the, what in the in the world of television, where we we collectively are still on a path to go from small data to big data, where have the Nielsen sample of tens of thousands of households being measured to big data sets, which are tens of millions.

Speaker: Right. And there are pros and cons of using both kinds. just The small data set, you can demographically weight and balance. Larger data sets are much more accurate because there's more data, but you have to make certain adjustments. So, yeah,

Speaker: It's at that time we were much, we collectively, the advertising marketing industry were earlier in the cycle of still relying on small data. I gotcha. Okay.

Speaker: So this was moving in the direction of getting more, more households, more real time across multiple channels and by channels, meaning, you know, different delivery services, but also the different actual channels.

Speaker: That's right. Yeah. Um, Okay, so on this topic then of different kinds of data from small to big, from occasional to real time, um towards the end of the book, you talk about the consumer price index and how it's collected and all the work that goes into that.

Speaker: And I'm curious if you have thoughts about where we are in the U.S., on, you know, I would say pretty dramatic changes to the federal statistical agency structure.

Speaker: um And we know that, for example, there the BLS is is cutting staff, that's, you know, whose job it is to create the CPI. And I wonder, I guess, from like, you know, whether you have any just thoughts on that generally, but also like,

Speaker: what do you think data back to your data person? Like what should your data person be thinking about in this kind of maybe new effort for for for folks who are relying on federal data for lots of different things? Like what should they be thinking about in their own day-to-day work?

Speaker: If you go back to the origins of demographics and you go back to the origins of data science, Data science really came alive in moments of life or death.

Speaker: the The first demographer slash data scientist is credited to be this fellow named John Grunt, G-R-A-U-N-T, who was a haberdasher in London in the 1640s.

Speaker: who, I don't know, he maybe he just spent a lot of time with measuring tape, but he but became he really fell in love with numbers. And he got really frustrated with the way the London authorities were dealing with the plague around London, because there were all these different neighborhoods in London. you know, London, when you go there now, when you go to the tube stops, you can see how they're all these kind of villages that were strung together. And at the time, it was the same thing, only more so.

Speaker: And John Grant was... concerned that all the data that was being gathered about causes of death was not being used by the authorities to help manage the plague and keep more people alive.

Speaker: And the technique at the time for gathering cause of death data was someone would die, they would ring the bell in the church, and what he calls ancient matrons, so old ladies,

Speaker: would sort of trundle over and they would make a note somehow of the cause of death that they observed or that they learned from the person who brought the body.

Speaker: And these were actually tallied up on a weekly basis and called the bills of mortality and they were published within the city. And John Grant took all the bills of mortality and tabled them up and made them consistent and did all the things you do with data to make it usable.

Speaker: And then he showed trends over time and you would see that someone died of of, ah you know, drowning in the bath at six people per year. And then people died of the plague is 60,000 people per year. And you would see it by neighborhood.

Speaker: And that's really when data science came alive, being using data to answer questions. Right. And. That was its birth. and And the other moment of this that i point out in the book is this Princeton stat statistician named John Tukey, who really foresaw a lot of what's happening now with the bridging of data and and computer science to answer big questions. And John Tukey was a Cold War and and Second World War ah statistician who

Speaker: helped make battlefield weapons for the United States government and helped use, create Cold War spy planes for the US government before settling into being a Princeton stats professor.

Speaker: And he too saw the the the relevance of using data from diverse sources and sometimes just good enough data, not great data to answer these life and death questions.

Speaker: All of just a long way around to say that there are There's so much tradition of data being public data being used to answer important questions.

Speaker: And I feel as ah as a data person to read about databases and data access being deprecated and being deprecated for what one can only guess are political reasons or ideological reasons is both heartbreaking for the endeavor of humanity to advance itself with knowledge and science, but also entirely against the grain of the tradition of data science, which to me goes back to John Grant and using data to save people's lives and public data to use data to save people's lives.

Speaker: Yeah. And do you... foresee a spot. Now yeah you've worked in the private sector, you know, different places. Do you foresee the private sector stepping in different ways with different data to fill in those gaps.

Speaker: I mean, not necessarily as a public service, but that those data will be the data that we will have to then rely on to, you know, maybe it won't be as ah comprehensive as the CPI, but we'll be able to track, you know, prices of ads on streaming television in a way that maybe the government is no longer able to do.

Speaker: You know, data businesses are not easy, even today when data appears to be cheap and plentiful. Most data businesses, you have to have a paying customer.

Speaker: And so having 60,000 scientists who are going to log into your system and download a couple of tables in order to answer an obscure question in their lab, to me, does not sound like a great customer base, unfortunately.

Speaker: And the other thing about the data business, again, even in this world of cheap and plentiful, is it's a fixed cost business. you you exite You exert a lot of effort up front. to gather data, clean it, figure out the use cases, make it available.

Speaker: And that upfront investment means you're putting a lot of money and effort in long before you're breaking even, much less making a profit.

Speaker: yeah So the the notion that entrepreneurs would be jumping into these gaps left by government cuts to resupply all the impoverished scientists with data, unfortunately seems unlikely for those reasons.

Speaker: Yeah. Well, ah yeah, I don't, I don't disagree with you and it will be interesting to see how, how things evolve over the next three, four years.

Speaker: um Okay. So the book, just to wrap the book is Little Book of Data. Yeah, by the way, yes, we completely went into a depressive hole there. And now we're going to pull up. butre We're going to pull up.

Speaker: We're going to pull Yeah, we're going pull Yeah, we ended up in a dark place there. so um So people should certainly check out the book. um They can get the book with the index and the author's notes, which would be ah helpful. um Where can people find you to to get you know get in touch with you know more requests, more information, you know workshops, whatever whatever it is? like where Where can they find you?

Speaker: My home base is LinkedIn. ah You can look look look me up, Justin Evans. I'm the one that says dad and author on the on ah my slug. It's got a big picture of the the little book of data as my background photo. ah But the little book of data is available in stores, including in airports. I'm delighted it's in Hudson News in the airport. Yeah, that's fun.

Speaker: And ah if you like the sound of my voice at all, you can hear me for four hours reading the book in an audio book form. Okay, terrific. I didn't know there was an audiobook for him. That's super fun.

Speaker: I, of course, have the paperback. I'm just going to keep it on the shelf here. This is great. Justin, thanks a lot for coming on the show. It was really fun to chat. And best of luck with the book. I hope ah i hope people will check it out. Appreciate it, John. Thank you so much.

Speaker: Thanks everyone for tuning into this week's episode. Hope you enjoyed that. Hope you will check out Justin's book, Little Book of Data. I will link to it in the show notes. And I hope you're enjoying this season of the show so far. I've got some really exciting interviews coming your way.

Speaker: And let me know if there are folks that you would like to hear from. If there are people, if there are organizations that you are paying attention to in your life ah that you would like me to try to reach out to and talk to, ah to see how they approach data, data visualization, presentations, tools, and so on and so forth, I will see if I can get them on the show.

Speaker: So until next time, this has been the Policy of His Podcast. Thanks so much for listening.