#75 The Data Storytelling Skills Data Teams Need with Andy Cotgreave, Technical Evangelist at Tableau

Hi, I'm Maggie, VP of Curriculum at datacamp, and I know firsthand how overwhelming it can be to keep up with the rapid changes in data and AI. But staying ahead doesn't have to be hard. That's why our curriculum team has curated the best learning content to help you and your team stay ahead, covering everything from data and AI literacy for busy executives to cutting edge generative AI for developers, and everything in between. Whether you're upskilling yourself or your team, we make learning practical, engaging and flexible. It just works. Power your data and AI transformation journey on datacamp.com.

Hello, this is Adele Neme from datacamp and welcome to Data Framed, a podcast covering all things data and its impact on organizations across the world. You know, oftentimes when we discuss the value data science brings to the table, a lot of the focus shifts to difficult to achieve machine learning and artificial intelligence. I'd argue though, that 80% of the value when doing data science comes from data driven decision making at scale. This is where business intelligence comes into the picture, equipping decision makers with the proper context and insight to make better decisions with data. This is why I'm excited to speak to Andy Cotgreeve on today's episode of Data Framed. Andy is the co author of the Big Book of Dashboards and his technical evangelist at Tableau. He is the host and of the if Data Could Talk podcast, co host of the Chart Chat podcast, and he's also a columnist for Information Age. He was on the 2021 Data IQ Top 100 Most Influential People in Data. With over 15 years of experience in the industry, he has inspired thousands of people with technical advice and ideas on how to identify trends and visual analytics and develop their own data discovery skills. And I had an awesome conversation with him today. Throughout the episode, Andy discusses his background, the skills every analyst should know to equip organizations with better data driven decision making, his best practices for data storytelling, how he thinks about data literacy and ways to spread it within the organization, the importance of community when creating a data driven organization, and more. Also, we'd absolutely love your feedback on how we can make Data Frame the better show for you and which guests you think we should bring onto the show along. I left the survey link in the episode description. Make sure to fill it out as I'd greatly appreciate it. Andy, it's great to have you on the show.

Thanks for having me on the show, Adele.

You're someone who has a very prolific career in the data space and are a fountain of knowledge when it comes to Data visualization and data storytelling. I'm excited to discuss all of that with you today, but I'd love to first start off by discussing your background. Being a technical evangelist at Tableau, I'm sure you get a lot of questions of what that entails and how did you end up here. So I'd love to first learn about your background, specifically what was the path you took that led you here?

So I'm technical evangelist at Tableau. I've been at Tableau for 10 years and I first downloaded Tableau in November 2007. The product didn't do very much back then, but it was still the best piece of software on the market. It blew my mind and it enabled completely new things that we were not able to do. So it's been an incredible 14 year journey. Prior to that, I was so. I was an analyst at the University of Oxford at that time, trying to track student data. And we just needed tools that were better. But how did I get there? Prior to that, I was working for a business research company, and at that company, my boss was always asking me questions about progress for my projects. And, you know, my colleagues was just giving reams of Word documents. And I was like, there must be a better way. And I ended up building this front page on an Excel spreadsheet which had a bunch of charts on it and summarized what was going on. I didn't even know they were called dashboards at that point. But, you know, I was like, well, here you go, Boss is one page. And somewhere along that process I learned that that's called dashboard. I was like, oh, that's quite interesting. It's a way of summarizing information. My earlier part of my career, you know, I'd left school wanting to go into art and do comics, but I ended up doing geography. Then I ended in computer science. I was a software engineer, I did business research, and I basically did this whole gamut of skills. Art, creativity, engineering, database, communication, design, that when I finally got to Tableau and really dug into what a data analyst should be, I realized that I'd somehow amassed all the skills that you need to really do this role to the full. You know, it's five years since I became Tableau's technical evangelist. And that role, that job, is essentially using my wisdom and expertise to bring that expertise and passion to the field to as many people as I can, and then hopefully bring them to Tableau, because that's my favorite platform of choice. Yeah. So that's the career summary in a nutshell.

That's awesome. And there's definitely a lot of wisdom to unpack here. So, you know, preparing for this interview was quite difficult because there's a lot of different angles by which we can approach this conversation. However, one thing I definitely wanted to pick your brains on is really the importance of data storytelling as well as the role of the data analyst in a modern data driven organization. I think we often talk about data scientists as the primary role for bringing value from data in an organization, but I'd argue scaling the amount of data driven decisions an organization makes is 80% of the value in the data pyramid. And this is something owned by analysts as much as the data scientists. With that context, I'd love to learn what you think are the main skills needed to be a successful data analyst today.

It's a really interesting question and my top answer is curiosity. Right? It's not even a technical skill. In order to be a data analyst, you've got to be able to hunt out what it is the data is telling you. You've got to be able to hunt out what it is your users are asking for and you've got to be able to explore and iterate to find the best articulation of those two things. What's the data telling you? How should it be communicated? So you know that curiosity feeds a sense of empathy as well, right? Because I could be a hardcore data nerd, but if I can't actually express that to the end to other people, express what I'm finding, my insights, then what's the point, right? You know, I think sometimes we can over focus on the technical skills and forget that. It's like, okay, you've, you've collected your data, you've prepared it, you've done something really wizzy technically. But can people see and understand what it is you found? And if they can't, then you missed it out. So curiosity and empathy are kind of the core skills, you know, underpinning that, you know, the balance of engineering and design skills are also really important. But I think we, maybe we can explore that a little bit deeper in a moment. But yeah, curiosity, I'm going to say 100%.

And you mentioned here technical skills, engineering skills, and I'm excited to unpack all of those with you. I'd love to expand on what you think are the technical skills needed for a successful data analyst today. And where do you think tableau fits in the picture in the wider tool ecosystem and analyst needs to know?

Fundamentally you need to get up to date on your database skills. So learning SQL clearly going to be a huge skill. When I started in this field, it was really only relational databases. So understanding data structures and SQL, fantastic. Obviously, since the world has developed, there's not unstructured data. There are many more different types of data models. So really getting in your JSON, your xml, all of that stuff is going to really help you to be a successful data analyst and data scientist. You, you don't have to be, you don't have to be super, super deep into those skills. You know, there are different paths you can take within the data analyst career, but you have to have a good foundation in that knowledge because otherwise none of what you, none of what you're going to do is going to, or you're not going to understand, you know, any of what you're going to do. So I think that's a real challenge right then. So that's kind of data structure, data architecture. You're going to do a lot of data preparation, cleaning and at that point, you know, your database structure is going to be great. But then you've got to think, well, what tools do I need to do this? Obviously we have tableau prep. That's, you know, I love using tableau prep. But again, getting your head around, you know, merging fields, processing data and you know, transformations, pivots, that kind of thing is also key. And then the higher or the more front end, customer end, you've got the analytics tools. So, you know, then it's, there are many tools in the market. Obviously tableau is amazing, but as you say, r Python, you know, these are all great skills to get into your portfolio for anybody who's really early in their career, you know, just start developing skills in one, one tool of choice. And as your career grows, you know, the pass of your career might steer you towards one or other newer tools.

Obviously technical skills are very important, but data visualization skills are also extremely important for data analysts. I've seen you speak about this, you've written a book on this and you've even extended this to say that data specialists should think like designers. I'd love it if you can expand that notion and what you think thinking like a designer means. And what are the design skills data data experts need to know to be successful in data visualization.

So this is something I'm so passionate about. And there's a woman called Kim Reese, she co founded Periscopic, which is a nonprofit data visualization consultancy in Oregon in the US and she sat on an episode of policyviz, which is another great podcast about five years ago. She said data visualization is a language, a means to convey an opinion or an argument. And that quote, I mean, that has stuck with me and sort of become my, you know, one of my North Star pieces of inspiration. Because in that it captures the fact that at our peril we think charts are truths. Like I have aggregated the data and I've made a bar chart and therefore that bar is bigger than that one. It's true, it's an uncontrovertible fact. It's like, no, because that aggregation and that visualization in a, in a certain way. And you can make hugely different, you can make charts tell hugely different stories with subtle choices of aggregations or design or color choice, orientation, the title, annotations, all of these things can change how the end users interpret the chart. And I find that absolutely fascinating. And the reason why design skills are so important is that you need to be able to wield that power deliberately. Right? You know, you could wield that power and frame arguments accidentally. You could even some people do it deceptive, deliberately deceive with the way they frame charts. So to be super powerful at this, you have to know when to turn these, some of these levers up or down to push opinionated thing or to try and be neutral and also to be able to critique other, other people's works and see when they might have inadvertently swung you one way or another and have that critique. So design is just absolutely fundamental to things, you know, think about dashboards. So, you know, I co wrote the Big book of dashboards, 28 real world case studies that we dissected and sort of described how they worked. And you're creating an object that is consumed by other people. That's what Apple does with iPhones, that's what Samsung does with TV remote controls. Even your kettle, even your washing machine. They are designed objects that have to satisfy various different levels of processing. A great book, another great book is Don Norman's the Design of Everyday Things. Another huge influence on me. And in that book, Don Norman says that any object that is designed is processed in three ways by every end user. There is a visceral level of processing, a behavioral and a reflective. That visceral is the millisecond, first millisecond of experience with that object where people judge it on how it looks, it's unescapable. So if your dashboard looks awful, you've already lost half your audience. So you've got to make something look good. You know, iPhones look great for a reason, right? You know, nice kettles look great when they're. When they're really nicely designed. And then the behavioral level is, well, okay, you've got this object. Can the person trying to use that object do the thing they want to do with it? Right. Or is it too complicated? You know, Don Norman has the example of doors in buildings that are designed to be pushed, but they have a pull handle on them. Right. And that's sort of a confusing user experience because it looks like a pull handle. And so the behaviorally, it's a bit confusing, but again, that's. You've got to put that into your dashboards. Can people actually work out how to interact with that dashboard to answer the questions? And finally after that, they will reflect and think, I did or I didn't like that dashboard. And if that reflection is successful, they will come back. So it's no good being an engineer and just thinking, the power of my engineering will succeed, because you've got to think like a designer to get that, the human experience to be a success too. So I could do an hour and a half on that conversation.

Definitely, we can expand on this. So you've mentioned here the power of design skills in conveying messages and it could be used deceptively. So we've especially seen the power of design on display in 2020 with COVID charts, climate change charts. So with that, what do you think? Are the weapons, to use that analogy, in the hands of designers that they can use to convey their message impactfully? What are the tactics that they can employ?

I mean, there's a really famous example from Simon Scar. This was a chart called Iraq's Bloody Toll came out in 2012. And it's a bar chart. It's data about deaths in Iraq from conflict. And what Simon Scar did was point the bars downwards and colour them a very evocative blood red and used the title Iraq's Bloody Toll at the top. So when you look at it, it looks like a smear of blood dripping down the screen. It's really emotive. And so what did he do? He used color and orientation. You know, should a bar chart point upwards or downwards? Well, normally it would go up, but he pointed them down and then he used the color to create a visual metaphor. But then also importantly, he gave it a title. Right. You know, just think about how reactive we are to titles these days. If you go to YouTube, you know, the wording of a YouTube video title is so clickbaity these days. You know, you won't believe the things, and I'm a sucker for that. I'm like, oh, better watch that. But, you know, research shows that the first thing people look at when they look at a visualization is the title. So you have this moment to grab people's attention and tell them what you want them to take from a chart. Right now that's extremely powerful. When I'm building dashboards or building PowerPoints that my CEO might see or he might, or she might even deliver, because the chart appears, everybody reads the slide, the chart title, and then there's the chart that proves the point, right? Then when you think about this in 2020, obviously coming through the pandemic. So that's an example of some of the techniques you have, color, orientation, size, all of that techniques. And what the pandemic has shown us is how carefully one must wield that power. So I'm based in the uk. Our government has had regular press conferences that have been, that have led with charts. You know, it's the, the Prime Minister would say some words and then he hands over to the chief scientific advisors who update the country, a nervous country, about this horrible pandemic. And you know, that team in the Cabinet Office in the UK had to work extremely hard to make sure the charts were neutral because.

You know, you.

Know, any, any perceived bias or influencing in that data visualization will get pounced upon by the opposition or by the media and they'll say, you know, they're trying to make us think X and Y and maybe they are right, but they can't be seen to be in press conferences. So what's been interesting in the UK is that experience of seeing the effort to reduce all the dials back to zero on anything that is not neutral or, you know, they're going, here's the facts, we're trying to express them in their opinion. And so that's kind of the spectrum of power. You can go from Iraq's bloody toll, which is full on in your face, you're going to think blood smearing down the screen to COVID pandemic charts. It's like we're just presenting the facts and there's a lot of uncertainty and we're making decisions based on need.

That's perfect. And it really depends on the situation, which is here at work context and what is the desired impact analysts wants based on their visualization analysis. Right.

It's not always, that's not universal. Right. Because if I'm building an exploratory dashboard that people can log on to and interact to track, you know, website visits or staff turnover, in that case, things have to be neutral because you're not really expressing an opinion about the data. It's just like here is an object you can use to track what's happening in the organization. But you know, a huge part of business is I'm going to go to my managers and convince them, based on evidence to make a decision to change the organization. Or I'm going to go to the colleagues or my own team and say, you know, we need to make a change. In that case, I am going to use persuasive charity because I've got an opinion and an argument to convey. So I can use, I can then in a business situation use that framing technique to sell my argument. And, you know, in a healthy data culture, that's as powerful as me using emotive words. You know, I think it's going to be a great idea if we put this podcast in front of a million viewers, right? You know, I could say those words and you can go, well, we can have an argument with words. We can do that with data too. So it is possible to do that in a business environment. But again, wielding the power without being deceptive. And also everybody having the fluency with data to know how to counter, how to process those arguments, to have a data driven conversation is powerful.

I'm excited to unpack data literacy with you. But before we do that, obviously data visualization skills, engineering skills, design skills are all super important. But the last mile of analytics is really the narrative you weave, as you said here, with words. This is what has often been called data storytelling. Can you share your thoughts and best practices on data storytelling and how analysts can gain better adoption of their solutions?

There's a couple of things that make my blood boil. One of them is when somebody copy and pastes a chart from one medium, so and then paste it into PowerPoint. The first thing that drives me to my head is when they don't even make it fill the screen, right? They just go copy, paste and they present it, right? I'm 50 years old. My eyes are not what they were when I was 20, right? And you know, this was a problem when we were in live meeting rooms. It's a problem now when we're looking at our laptops and these things. And when somebody says, if somebody says, as you can see on this chart, blah, blah, blah. If I can't see that, it's like you have failed in your job as a storyteller because I can't see your blooming chart. And you know, so often I've been in presentations where A chart comes on screen and so they say, oh, it's a bit small on the screen. And I'm like, if you have to say that you are lazy because you have not thought about the storytelling power of actually using this data to drive change. So that just drives me potty. So if you do nothing else with any chart you have on a PowerPoint slide, at least make it fit the screen, right? Secondly, it is when they say, oh, this chart's a bit complicated, but it shows, blah, blah, blah, blah. And I'm sitting there going, you've literally copy and pasted a dashboard onto a PowerPoint. It's got 10 charts on it and, you know, 96 pieces of information. How the hell am I supposed to know which bit of that you're telling you're referring to? You know, your word. The words coming out of your mouth might be, it shows that sales are down and we need to do X. And I'm just looking at a mess of information. I have no idea what I'm doing. The reason these make my blood boil is because as a storyteller, if I'm going to stand in a. If I'm going to convey information in a meeting, in a presentation, in a keynote, in anything or an email, how does the end user know what they're supposed to be looking at on that screen at the time you look at it, right? So thinking about storytelling is what is your narrative through any visual you put on a screen? If you're going to spend. If you're going to put a chart on a screen and that slide's only going to be visible for three seconds, then that chart has to be understandable in about one and a half seconds. If it is, if it can't be, it's the wrong chart. If you really want to show a complicated chart that takes two minutes to explain it, spend two minutes explaining it and bring each element of the chart forward one at a time. You know, it's like, this is what the axis is, this is what the X axis is. Each mark represents a country, and the color represents the continent and the size represents population. And then you're bringing the audience along so that you can actually explain the point. If anybody wants to see this done the best way possible, go check out Hans Rosling's TED Talks, because he was the master at this. And, you know, I mean, there are loads of different techniques you can. There's loads of different techniques, but storytelling is absolutely vital because you're using data to try and drive change in your organization by sharing insights Every single visual you show has to be part of the narrative flow of that conversation point. So yeah, sorry, don't make my blood boil. If you ever do a presentation and say to me this chart's a bit complicated. But then I'm like, ah, you've let.

Me down in terms of driving that narrative. I think something that data scientists struggle with is fitting a narrative and finding that narrative structure to their data or to their exploration. Is this something that can be templatized or is this a highly specific ad hoc problem that needs a highly specific ad hoc solution?

Yes and no. Right. It's difficult to say I can't tell anybody what the right narrative process is for the thing they're doing because it all depends on what their specific process is. Right. But that's not very helpful to somebody learning out, is it? So my good friend Ben Jones, he used to be at Tableau and now runs dataliteracy.com he did a great set of content about the seven different types of story. And anybody who's familiar with fiction might know there are seven different types of story. The quest. Yeah, right. And he sort of did some work to think, how could you tell different data stories based on. On different approaches? Do you start out with a big picture and zoom in? Do you start zoomed into a really low level piece of detail and then zoom out? Do you show progression over time? Do you showing progression over time? Can you bring drama about one data point through this? So there are ways in which you can think of a story. They will often lend themselves to presentations or narratives in email. If you're doing reports, what is the narrative through a report in dashboards? Story. Story is slightly strange in dashboards because the dashboards are largely exploratory facets. I'm going to go to the company sales dashboard to see what's happening in sales. But for me, a lot of the inspiration for dashboards comes essentially from comic books. I left school wanting to be a comic artist when I was 18. And a lot of the, the learnings from the world of comics is how they use a layout on a page to create a flow of time and direction. And I mean, if you get into the study of comics, it is amazing seeing the different ways they can create narrative flows with that page and frame structure, you know, so on a dashboard you're essentially going, you know, the top left is where people are going to look at first and then you can largely follow a left to right, down a bit, left to right structure. The Gutenberg diagram is another design Principle from print about how people look at pages. So you kind of want your most important thing in the top left and then more granular levels of detail towards the bottom, towards the right. It's not universal though, because you can play around with that kind of stuff.

Given that we just discussed the data visualization and design skills needed to be a successful data analyst, I'd love to pivot more to discuss how this looks like from an organizational point of obviously bridging the gap between reality and what's possible in terms of making use of data at scale is very important for organizations. However, I argue that data literacy is the biggest obstacle to that value being realized. And we can discuss that. I'd love to hear your thoughts on how you define data literacy and what it means to be a data literate organization for you.

So this is really important and I think there's various, there's various levels of data literacy. I think largely as a definition, being data literate is being able to do some or all of read, consume and critique. Read, consume, critique and create visualizations. Right. Not everybody needs to be able to do that. I would say everybody in an organization needs to be able to be comfortable reading, you know, looking at and understanding what charts are showing. And it's not a hard skill, but a lot of people just have not been taught that skill. You know, the subtleties of the way color can be used to influence things, which is the appropriate type of chart and is an important skill to learn to read. So being able to read and consume those, but also being able to critique them. You know, there's a lot of research going that shows that people see charts as truths. Right. It's hard to criticize a chart. Maybe I could criticize your paragraph of text, but you put it in a chart, so it must be true. So yeah, it's absolutely. So being able to look at these and have that critical mindset that, you know, you don't need to mistrust things, but you need to be able to question them and then produce charts as well. Should every single person in the organization be able to use tools to produce visualizations? I mean, ideally I'd say yes, but realistically probably not, but yeah, so, so that, that they're the four key pillars. And how do you test if your organization, where your organization is in that ability? It's, it's really hard for people to self admit that they are not data literate. I mean, you know, you know, I don't have a problem with the term data literacy, but some people, some People I think legitimately do because it's like, are you telling me I'm illiterate in this, in this thing? It's like, you know, so there's negative connotation to being illiterate. So, you know, another phrase could be data fluency, because fluency sort of has a baseline which is, I'm not very fluent and I'm really fluid, but I can still be at the baseline. But the reason I raise that is because it's actually quite hard to assess your organization's baseline. You know, you can do, you can survey users, you can ask them to self assess, which is useful and that can create a benchmark which you can then change. Yeah, so I think that's a definition of data literacy. Trying to benchmark your organization is key. And then the next step is how do you improve that fluency? What we found at Tableau over many years is that one of the great things is creating a community. Tableau has this crazily excited and excitable community of analysts and geeks who just love using tableau and love the field of analytics and the field of communication and engineering. And, and we have seen that that generates a momentum in and of itself. We often say that our tableau community won't let you fail. There's when people start getting involved, they ask questions and those forums and Twitter and social media video series, people will offer help. And there are many, many solutions out there. And what we reckon that what we now see is when customers try and enable an internal community, then that is a really good way of creating a virtuous circle that builds its own momentum. You know, one example, something I do at Tableau, we do a, it's called, we call Invis Club. And what we do is we find a chart from the world and once a month and we just sit around and critique it. And the reason I run that is because it's a super low level barrier to entry.

Right.

Anybody can come in and say whether they what they like or dislike about a chart. And it's really not intimidating for people who are lower down on that fluency track. You know, I could come in, I could make do a club and say, hey, it's the monthly let's hack tableau club. And instantly I'm going to eliminate 90% of people because they're like, oh, not very good at tableau yet. Or I'm not so comfortable with analytics. And so it's really high barrier of entry and quite intimidating thing. So you can take really small baby steps. You know, formally Training strategies, You know, build what is your data literacy program? How are you going to fulfill, improve those four criteria? And there are plenty of training materials. You know, we have a free data literacy course. Obviously Data Camp has loads of things. So yeah, it's about building that strategy to improve everybody's level.

That's really interesting when you mention the community building aspect of it. I think a lot of organizations sometimes forget the aspect of building a learning ecosystem rather than just, you know, dumping training courses onto their workforce. This involves live trainings, meetups, expert talks, hackathons. And we found that works best even on our side with Data Camp customers, especially when creating a data culture. If you were to design a data literacy program for an organization, what would it look like? Of course this is highly Persona specific. So if you want to talk about, you know, the lower end of the data fluency spectrum, would you focus on a tools agnostic conceptual program such as, you know, understanding use cases of data, or would you have a tool specific training in mind?

I'd start tool agnostic. I think, you know, for the people who aren't producing charts, you know, they, for them it's, it's how do they learn to read, right. And you know, they will have been taught hopefully basic skills at school, but not some of the more subtle techniques. So it's like charts work. Fundamental of data visualization is that the individual atoms that make up a chart, like a bar or a line or a dot, those things are called preattentive attributes. That's a bit of a generalization, right? But the concept of length or position or angle are things that we cognitively see instantaneously when we, when we look at something in front of us. And it's those preattentive attributes that enable us to make visualizations. Once you know, things like that, it suddenly means now I know the individual pieces, I can actually understand how a chart is put together. And then once you understand that, that sort of super granular level of how a chart works, it makes it easier to consume it. So I think teaching those fundamentals is really important and it just allows you to then, you know, build up to words. I mean, it is, it's like learning to read. You learn, at first you learn letters, but then simultaneously you're learning letters, but you're also learning words, you know, short words at the start and then at the same time, you know, and then simultaneously you're beginning to get sentence structure. So that's how we teach kids to read. We kind of do this hodgepodge of the Individual bits, the words and the sentences. And then. Right, so. And then as you mature, it becomes writing essays, but also reading books, you know, and then eventually you get to university and not only you're reading books, you're writing books, but you're deeply thinking about literacy and communication. And that's essentially the same framework for a data literacy program. How do you learn the absolute fundamentals simultaneously with practice, put some of them together, maybe run a little bit before you can walk, and then eventually just add more and more depth? So I think taking that model to your own strategy is great. At some point, soon after the fundamental bit, then obviously you've got to become a practitioner, at which point tools become important.

Couldn't agree more. And here, circling back to something that you've also mentioned, I've seen you speak about extending data literacy programs to schools in order to prepare future generations to work and think with data. Do you expand. Do you mind expanding on some of your thoughts here?

Yeah. I've got two daughters, teenagers now, and it's been quite funny. They have slightly more advanced data skills than their peers because Daddy's taught them or bored them, bored them a lot with all this data stuff. But it is interesting watching them go through the experience of the UK education curriculum, where they're encouraged to make 3D exploded pie charts because they're funky, right? And you think, ah, so they actually. They have a relatively strong grounding of what. Well, they have a grounding of what basic data is. Right. And. And the visualization is a thing, but there's nothing about why it works. Right. They're just told, make some numbers, build this thing. So there's none of that. There's none of that foundational knowledge. And I think that's the thing that's missing when you understand, you know, it's always more. I find it more fascinating to know why something works than just do it. Right. So if we could change the education curriculum to be a bit more about why these things work and then, you know, as we just. In the previous answer, if we can get them more to the how to critique and be a consumer of this information, then I think it's a really important thing to do at school because, you know, it's 20, 21. We've just had a pandemic. We're in a world of misinformation. You know, it's a scary time for kids coming into the world. And are we equipping them to be able to better interpret the world around them? The avalanche of information that's coming at them. I don't think we are. We're just telling them to make 3D pie charts when instead we could be building fluent data literate kids.

Especially when you think about it from a future workforce readiness perspective, I think that a lot of organizations are doing the heavy lifting now and preparing their workforces for the future. There needs to be modernization of the skills taught at the school level as well.

Now at Tableau we've been doing, I mean we've got a great academic program which helps students, but we've also been doing a data for kids program as well, trying to come up with funky exercises and you know, interesting exercises to get kids thinking about data literacy. I, I certainly did a few during the pandemic when we were homeschooling. And it's really amazing how you can get them to come at this world of data without, without them going, oh, dad's talking about data again. Oh, you know, you can bring it to their lives without them quite realizing what you've done.

That's really wonderful. And we've also seen a lot of good success in data camp for classrooms and for colleges. I'd like to circle back to dashboards here and the role they play in democratization of data insights and maybe hear play devil's advocate a bit. How do you view the role of dashboards in enabling organization data literacy? Similarly, what do you think are some of the drawbacks of dashboards? Do you think that there is a risk of having too many dashboards and information glut and this can hurt, hurt an organization's ability to make decisions with data and its entire data literacy program?

You can bash the dashboards all you want. I mean, I think one of the biggest, one of the potential downsides of having written a book, the big book of Dashboards about dashboards is that people then think, I am an advocate that dashboards are the be all and end all of your BI strategy and nothing could be further than the truth. I also do a whole presentation, a bit of content. There's a chapter in the book about don't build a dead end dashboard. You know, they are the start of the journey, not the end of the journey. Right? So dashboards are amazing, right? Every organization is going to is is asking a question about their organization or their customers. Do they know they're going to ask every week or every month or even every year? Right? You are going to monitor things. Is the website up or down? Is a sales tracking? Is turnover getting dangerously high or is turnover great? Right. Is stuff turnover good or bad? That questions, we want to know those. Every month we're going to track them. So whatever you call it, whatever the vehicle is, you need some sort of monitoring display so that people can keep an eye on these key performance indicators, or OKRs, whatever you want to call them. That said, if that's all your business has, oh my God, you failed completely. Completely. Because, you know, I mean, think in your own organization. Do you think in the next seven days somebody is going to ask a question that you hadn't anticipated? And if your answer to that question is no, then, I mean, I'd love to see your organization, every organization is dealing with uncertainty, resilience and ad hoc things, right? I mean, the pandemic turned everybody upside down. And it was organizations that were really digitally strong had great resilience because they were able to respond quickly to things they hadn't anticipated. So at that point, a dashboard's useless. A dashboard is going to answer. A dashboard answers three or four questions that you asked years ago, right? What about the question today? So you need this fluid, messy, exploratory, iterative, rapid response approach to data too. And that, that requires strong data sources, skilled users and literate people who can, you know, data fluent people who can ask the questions and respond to what they're seeing very, very quickly to get to the ad hoc answers. So yeah, dashboards, holy cow. Dashboards are a part of successful analytical strategy. You are failing. If you only build dashboards, can you build too many? Yeah, that is also a big problem. And you know, we see customers, they're like, we wanted to escape Excel hell and now we're in dashboard hell. It's like, you know, that's, that's not great, but again, great processes. And this is where actually so architect, not data architecture, but being able to architect a good business intelligence stack and process is really important. What we do at Tableau is we have anything that gets published has an expiry date and you know, as that's getting closer, we'll start getting notifications, then emails and being like it's going to get deleted and nobody's been looking at it. So we, we sunset a lot of work because we recognize dashboards die, right? They do. Dashboards die. It's ongoing creative destruction continually is really part of the strategy to keep things fresh.

That's great. I want to harp on here the importance of resilience and agility. I think not a lot of people discuss this as being a part of a data driven organization, but it's very important to kind of instill that culture of resilience and agility. So what are the ways that you think you can forge the sense of resilience?

Success in the data culture, there's three things. You need to have an agile system. So that's about creating that architecture that has the robustness of data sources, but also the ability for people to find things and respond quickly. There has to be a high level proficiency. So that's the skill base across an organization. The data fluency, but also the hard technical skills of using the platform you've implemented. And then the community is the third aspect of success. And you know, that's about having people who are engaged. In order for all of that to succeed, you also need buy in from the top. We've seen, we see customers that succeed when the executives are in on this too. They don't have to be making dashboards to be bought in, but they have to be using the data and really buy in. You know, we see a lot of failure where something grows organically in a small part of the organization, but they can't. And they get kind of, yeah, yeah, keep going from the top, but they don't get that organizational buy in. It has to become a data has to be a strategic asset in order to drive that success. So again, it's data literacy, data fluency, ensuring you've got the engineering skills and the architecture installed, the BI strategy to allow flexibility and rapid response. And it's hard. It's really hard.

Yeah. That's awesome. Especially executive buy in and leading by example. They're so important now pivoting to discuss the future. Where do you view the role of AI and automation coming into play within organizations? You know, there's often a lot of discussion around intelligence, augmentation and you know, augmented analytics. Where do you fall in the automation versus augmentation debates? What are some of the key trends you're excited about in the BI and AI space?

I'm all in favor of AI that augments a user and gets them from A to B quicker. I think most certainly the BI industry is investing heavily in that and Tableau is too. For example, we've got Ask Data and Explain Data are two new technologies that. Well, actually they're many years old now, but they're really iterating fast. Ask Data, for example, is a natural language interface to a data set that in its latest iteration. I'm really excited about it because what I fell in love with in Tableau 14 years ago was this ability to drag and drop things around this interface and just be in Complete control of the data. But you had to have an analyst's mindset and an analyst skill set to really embrace that. And hundreds of thousands, millions of people have done. But if you're not a data nerd, who knows the difference between a discrete and a continuous data field, that's a little bit intimidating. Ask Data now lets you type in a question, a bit like typing in something into Google, and you get an instant response. But what I love about what we've done with it is that it, as you type the sentence, it becomes something you can click on and iterate to change each part of the phrase of what you've written. And for me, it's the perfect extension of the power the analysts have had putting into the hands of, you know, non analysts, just normal business workers. And I had, I, I don't know if it's an epiphany, but recently I realized that, you know, the end goal of this is that there's always going to be a space for analysts in the tableau world. Right? Because there's always going to be more than this. But eventually you'll get to the point where people stop being excited about using Tableau, right? Or whatever, or any other tool that gets there. I'm like, what Andy Cockreave saying, I want a world where people are not excited to use Tableau. Huh? Let me explain. I use Microsoft Word every day to write my blogs and my content and I use PowerPoint and Google Slides. I don't come out at the end of the day and go, oh, I had a great day using Microsoft Word. I just come out at the end of the day going, I wrote some great content. Right. You know, because that's what you do. Because the tool, the tool just does its job. You know, I don't, Right. And so with Ask Data, as we put data in the hands of more people, they're not going to be, they're not that, hopefully. You know, I think the great goal is they have been augmented to the point that they don't have to think about the tool they're using. They're just asking, hey, I asked some questions about my data today. And AI machine learning really, really help empower that, you know, and again, our integrations with Slack are going to improve to the point that you can do this all in Slack. And that's driven by AI machine learning. And yeah, to the point that people stop thinking they're using Tableau. They're just asking questions of data. And that's great, you know, so, yeah, I think I'm saying I'm campaigning for a world where people are no longer excited about using Tableau. Yeah, that somehow seems strange, but hopefully people know what I mean.

Definitely see where you're coming from, especially when there's this kind of ready to use infrastructure to support data driven decision making through a lot of these tools that you're discussing. So given that, I also agree that there's always a place for a data analyst and an augmented data analyst, as you say, what do you think is the most useful skill that analysts need to know in a world where a lot of simple data workflows are either automated or semi automated?

Well, I think another good metaphor is Instagram, right. Mobile phones, you know, mobile phones have made us all professional photographers and we can do, you know, we can do filtering techniques that traditionally would have been done by true masters, you know, true experts in the developing lab. Right. They could, they would apply things as they process photos that we can now do in the space of a click. And in some ways that's again the goal. Right, but, but there is still a space for professional photographers, right? You know, you know, so that's great. That is still going to be the case for analysts. There will always be the unanticipated sides of things. There will always be, I suspect there will always be edge cases that natural language just won't be able to query. Right. So I think the life of the analyst is not under threat in terms of skill sets, if I'm really honest, I'm not sure I have a really solid prediction of what the next skill sets are going to be. Because the reality of what I see is a BI industry creating really advanced technology that promises complete revolution. When I go to customers, 90% of them or thereabouts are still using PDF and Excel data sources. Right. And so, and they're looking at this AI stuff and going, I still have to copy and paste the table for this PDF into a CSV file and then connect it. So much as the cutting edge of the industry is, wow, it's sci fi, actually most customers, it's going to take decades for them to chase. So I think the traditional skills are still going to be valid for a significant time in the future.

Awesome. Finally, Andy, as we wrap up on an inspirational note, do you have any final call to action for today?

If anything, if anyone's listening to this, I would just say find a community and join it. Examples we have Tableau Public is our free place where you can share visualizations. And we've got things like Workout Wednesday, viz for social good. All these social projects where people hammer out and play around with data sets and share their findings together. And it's just, it's really enriching because joining that community, it exposes yourself to other people. It puts you in a learning mode straight away, but it also puts you in a teaching mode straight away because anytime you share some work you've done, you're going, well, this is my interpretation of the task. And you know, something we've learned many times over the years about critiquing in a constructive way is really important. So being able to hear critique, but add critique and, you know, in with the motivation of elevating everybody else is hugely beneficial, particularly on Tableau Public, but others as well. You will also build a portfolio that shows the progression of your skill. And it's slightly strange that from an engineer's perspective, it's like, I don't need to have a portfolio of my work, but from the design perspective, graphic designers have portfolios. So, yeah, join a community.

Thank you so much, Andy. Really appreciate you coming on Dataframe and sharing your insights.

It's my absolute pleasure. Thanks for having me.

That's it for today's episode of Dataframe. Thanks for being with us. I really enjoyed Andy's insights on visual analytics and data storytelling best practices. If you enjoyed this podcast, make sure to leave a review on itunes. Our next episode will be with Vishnu Ram, VP of Data Science and Engineering at Credit Karma. I hope it'll be useful for you and catch you next time on Data Framed.