Adel Nehme, the host of DataFramed, the DataCamp podcast, recently interviewed Brent Dykes, Senior Director of Insights and Data Storytelling at Blast Analytics. He is also the author of Effective Data Storytelling: How to Drive Change with Data, Narrative, and Visuals.
Adel Nehme: Hello, this is Adel Nehme from DataCamp and welcome to DataFramed. A podcast covering all things data and its impact on organizations across the world. Something we mentioned earlier in year in our 2021 data trends report, is the rise of data storytelling and how it's not going away any time soon. Data storytelling is often called the last mile of analytics, as it's the final hurdle before data solutions are adopted within the organization. This is why I'm excited to chat with Brent Dykes on today's episode. Brent Dykes is the Senior Director of Insights and Data Storytelling at Blast Analytics. He's also the author of Effective Data Storytelling: How to Drive Change with Data, Narrative and Visuals. Brent has more than 17 years of enterprise analytics experience at Omniture, Adobe, and Domo. His passion for data strategy and data storytelling comes from consulting with many industry leaders, including Nike, Microsoft, Sony, and Comcast.
Adel Nehme: He is a regular Forbes contributor and has written more than 40 articles on different data related topics. In 2016. Brent received the Most Influential Industry Contributor award from the Digital Analytics Association. He is a popular speaker at conferences, such as Strata, Web Summit, Adtech, Adobe Summit and more. Brent holds an MBA from Brigham Young University and a BBA in marketing from Simon Fraser University. Throughout the episode, Brent and I talk about his background, what made him write the book on effective data storytelling, how data storytelling is often misinterpreted and misused, the psychology of storytelling and how humans are shaped to resonate with it, the role of empathy when creating data stories, the blueprint of a successful data story, what data scientists can do to become data storytellers, the future of augmented analytics and data storytelling and more.
Adel Nehme: If you enjoyed today's conversation with Brent, make sure to also register for upcoming webinars with him, where he'll be visually introducing us to many of the concepts discussed in today's episode. The link is in the show notes. If you want to check out previous episodes of the podcast and show notes, make sure to go to www.datacamp.com/community/podcast.
Adel Nehme: Brent, it's great to have you on the show. As many listeners know, and this is something that you've mentioned in your book, working with data and presenting data insights is an integral part of the present and future work. And I think your book on effective data storytelling is one of the best out there. Before we deep dive into effective data storytelling, can you give us a brief background about yourself and what led you to write the book on it?
Brent Dykes: Yeah, absolutely. So my background is in marketing analytics, specifically digital analytics. I joined as an analytics startup called Omniture, and then we were about four years into that, we were acquired by Adobe. And so I spent eight years in consulting working with fortune 1,000 companies. And then four years, I was an evangelist for Adobe analytics platform. And then after that, a former manager recruited me over to Domo. And if you've heard of Domo, they're a cloud-based BI solution and I was there for four years. And then in 2020, I joined Blast Analytics and I was establishing a data storytelling practice. And so getting into why I wrote the book, back when I was at Adobe, I saw the challenges that a lot of data professionals would face in sharing their insights effectively. And so I pitched the concept of doing a breakout session on data storytelling back in 2014, which would have been a part of Adobe's summit, which is the customer conference that they have.
Brent Dykes: And the session was a big hit. We had a great attendance, a lot of interest and that was my first signal that maybe I was onto something. And then I started to deliver at different conferences. And in a lot of times, people would come up to me after my session, say, "Great session, Brent. Do you have a book?" And I'd say, "No, I don't have a book or at least not a book on this topic." And then another signal came when I wrote an article. I'm a Forbes contributor and I've written lots of articles on data related topics, but my all time most popular article has been one where I focused on data storytelling being an essential data science skill that everyone must master. And so I think when I had all those signals, I was like, okay, I got to write a book on this. Because it's something honestly I'm super passionate about, and it's something that I think there's a real need out there. And I think we're just seeing more and more that need today, even 10 years ago, I think it's got even more important today than ever.
Adel Nehme: That's great, I'm really excited to discuss the ins and outs of data storytelling with you. I think that the term data storytelling is often obfuscated as it become a buzzword or it's in a hype cycle, especially in data space. Do you mind sharing your thoughts on how you think the term can be misused or misinterpreted and what you think data storytelling should be?
Brent Dykes: Yeah. That was another reason why I wrote my book because I was seeing the term being misused in the media and also by analytics vendors. Obviously it's a sexy term. I think when people talk about storytelling and then you latch on the term data on there, lots of people started grabbing onto it without really understanding what it is or even what it isn't. And so three misconceptions that I've seen that have bubbled up, the first one being that data storytelling is just a synonym for data visualization. Definitely data visualization is a core element of data storytelling. Often we're dealing with complex data sets and we need to share the information in a way that people that aren't as familiar with the data can comprehend and follow, but I really do see it as a means to an end. We're trying to tell a story with the data. We're trying to convince somebody to take an action or do something. And so the visualizations are just a means for getting people to understand our insights.
Brent Dykes: And I would even say that we don't always need to have visualizations as part of a data story. I mean, I'm sure there's lots of you, many of your listeners have probably heard great audio only podcasts, talk about facts and figures and tell really compelling data stories. So I don't think that visualizations, we can't look at data storytelling as similar for data visualization work. So that's the first misconception I've seen. Another second misconception that I've seen is that dashboards tell data stories. I even worked for a vendor who would say that, "Oh, we'll help you to tell stories with your data using our dashboards." And that was something that I struggled with because, if we're talking about an automated dashboard, you're building your dashboard, you're obviously pinpointing KPIs or key metrics that are really important, so you put those in the dashboard.
Brent Dykes: But when I'm building that dashboard, I have no clue what's going to happen with that data. I don't know if it's going to spike up, spike down, completely flat line forever. I don't know what the stories are that are going to come out of that dashboard. And so I think the key thing about a dashboard is it's an exploratory tool really. I mean, it's more geared towards helping to share and disseminate information, but dashboards aren't designed to really communicate a specific insight in the data. And so I think that's another misconception that I saw.
Brent Dykes: And then the third misconception that I see often is that, oh, we just put, for the narrative part to turn something into a data story, oh, we just slap some texts in there to go along with the visualization, oh, we're telling a story now? And I'm not bashing on annotations in commentary, those are very important to storytelling, but that's only a piece of storytelling. I think we need to broaden our perspective to think about, in terms of the narrative structure, the flow of how we're sharing information, how we organize and how we emphasize things in our findings and with our observations and insights, that's really how storytelling comes together. And there's lots of facets to that, but it's not just about some texts with the data visualization. That's not storytelling by itself.
How can data stories be used as a device to create organizational change?
Adel Nehme: So I'm excited to define what data storytelling is with you. But first I'd like to uncover the value of data storytelling and why it's so important. So one of the key notions mentioned throughout the book is how data storytelling can be used as an agent or tool for change. Often when organizations or people confronted with data that challenges the status quo of some sorts. There is natural resistance to change. I'm sure many data scientists listening in have faced some of these resistance when talking with business units, business executives. How can data stories be used as a device to create organizational change?
Brent Dykes: Yeah. I think one of the key things that I think we overlook often is that what is an insight? An insight represents a potential change. I mean, an insight invite change. We're basically saying, stop doing this and do more of this, or let's invest less money here and invest more over here. Basically, when we get an insight. And one of the things, when I was writing my book, a reviewer mentioned, you talk a lot about insights are you're going to define what an insight is. And that caught me and I was like, oh, I wonder what an insight actually is. And so I started looking at definitions and there's a lot of useless dictionary definitions of insight. And so I was finding lots of those and they weren't that helpful. And then I came across one from a psychologist by the name of Gary Klein. And he defined it in his book a, it's an unexpected shift in our understanding.
Brent Dykes: And for me, that really embodied what an insight is for me. Because it catches our attention because it shakes our understanding of the world as it is. And then what do we want to do? Like if it's in our ability to act on that insight, because it's really just an insight for ourselves or personal insight maybe for our role at our company, we can act on it ourselves and we don't need to tell a data story. However, when it's a situation where, okay, I've got an insight, it could change how our team operates, or what our company's going to do, or department, or whatever it is, I'm going to have to get buy in. I'm going to have to get maybe resources, budget, support to make something happen. And that's when we need to tell data stories, because we're basically sharing an insight that's going to drive a change.
Brent Dykes: And as we know, people are always going to be resistant to change. And so that's why communicating our insights effectively is so critical. And the key thing here, we've done a lot of work to find an insight, all of the capturing of the data processing of the data, organizing, analyze it, we build a model, and then we come away with this insight. All of that work will be for not if we're unable to communicate it clearly to other people and have them consider it and act on it. And so it's like, often we talk about it being the last mile of analytics that, we do all this work and then if we... And I've seen this happen all the time where an insight is not communicated clearly and all of that work that preceded it, is undone.
Brent Dykes: And so I think, a lot of data scientists and analytics professionals, they're realizing that I can't squander all of my hard work by simply not communicating my ideas clearly and in an effective manner. So that's why data storytelling is going to help with organizational change, because if we can communicate our ideas and our insights clearly, it's going to help the organization to embrace that change than before.
Adel Nehme: Couldn't agree more. And in the second and third chapter of your book, you really deep dive into not only just the value of storytelling, but the science and the psychology behind storytelling and how it resonates with humans. Do you mind sharing some of your thinking around how data storytelling resonates with people in influencing decision-making and creating organizational change?
Brent Dykes: Yeah, no. I think it is a key factor in influencing decision-making and one of the key things is that I don't think we fully appreciate how much we rely on storytelling in our daily lives. It's deeply seated in who we are as human beings. I mean, from all of the books that we read to all of the movies and the media, and then even, as we interact with colleagues at work, we're sharing stories, what happened on the weekend, or why isn't that team working well? Oh, let me tell you why. Or different things, I mean, with our family, with our kids sharing stories, oh, that sucks what happened to you with your friend there. Let me tell you, I had something similar happening with one of my friend, and we're constantly storytelling.
Brent Dykes: Even when we turn out the lights, our brain doesn't shut off. It then starts to formulate stories in our dreams. And so it's a big part of who we are as human beings and even going back, there was a study done in the early 70s where researchers studied one of the last nomadic tribes in Botswana, and they discovered that 80% of the campfire discussions centered around storytelling. And that's really how they taught the next generation like, oh, when you're looking at that plant when it's yellow, don't eat it. Wait until it turns purple, then it's going to be edible and not poisonous. And they'll tell the story of that. And that's really how we form a community, how we connect with other people. And there's two reasons why stories beat statistics and they are because two things, one, because they're more memorable and two, because they're more persuasive.
Brent Dykes: And there's a couple of examples of this today that I borrowed from Chip and Dan Heath's book Made to Stick. I don't know if you're familiar with that excellent book on communication, but in one of their example... Chip Heath is actually a professor at Stanford University and he actually teaches a communications class and he puts them through an exercise where he gives the students a bunch of data points around a particular topic. He asked them to take a position on it and then create a short presentation where they incorporate these data points into their presentation. Now, 1 in 10 of these students will actually also include an anecdote, or a short story, or something that they weave into their presentations. And so the students present to each other, they rate each other and how they did, give feedback. And then they think that the exercise is over.
Brent Dykes: But about 10 minutes later, Chip Heath comes back to them and says, "Okay, how many of you remember any of the statistics?" And only about 5% of them could remember any of the statistics, but in terms of those stories that were shared, 63% of them could remember those stories. And so that's just an example of how stories can be more memorable than statistics. Now, on the persuasion side, there's another story that they share in the book that I also share in my book, which comes from Carnegie Mellon University, where they did a study with a bunch of students, they had them take a technology survey, the students complete that survey and then that's where the experiment begins because they give them five $1 bills and they say, "Hey, here's this charity," it's a real charity, Save The Children. And they had a brochure for this and they shared the brochure with them and now there's two versions of the brochure.
Brent Dykes: One version of the brochure had a bunch of data around the suffering of children in Africa, due to famine, or war, or illnesses in different things, a lot of statistics and data. And then another version of that just talked about Rukia, seven-year-old girl from the Mali, talked about the suffering of her family. And so they gave them this brochure, these two versions, and they asked them, "Would you like to donate to this charity from the $5 that we've given you?" And the story version generated more than double that of this statistical version, even though we're talking about just one person in her family, it was relatable, it connected with these people and it encouraged them to donate at a much higher rate. So storytelling is super powerful. We need to tap into it so we can, get the memorability and the persuasion benefits.
Adel Nehme: I think this also speaks to the Testament of the power of empathy and really zeroing in as well on the concept of how stories meet statistics. One of the things that you mentioned throughout the book is how, and this is something I think data scientists will find a bit counter intuitive since they tend to be more analytical and much more evidence-based in their discussions. Do you mind highlighting the big differences, especially when it comes to the psychology of storytelling between how people are geared to react to facts versus how they are geared to react to stories?
Brent Dykes: Yeah. I mean, when I started in my career, I didn't realize how much, I thought that if I had the logic, I have the facts, I had the data, bring that to a decision maker. They're going to make a decision, that's logical and well-reasoned. Well, as you get into your career, you realize, man, a lot of irrational and disappointing decisions are being made, not based on data. What has it being based on? Well, it's been based on emotion. And I think I underappreciated earlier in my career, how much emotion plays a role in our decision-making. And it's interesting that, if we look at how people react to facts, one of the things they do when you approach somebody with facts, they go on guard, they're more skeptical, they don't want to be deceived or tricked by the data, however, if we approach them with a story, all of a sudden that intellectual guard goes down and people become much more open-minded, they're not going to nitpick on the details as much because they want to see where the story is going.
Brent Dykes: And even some psychologists have found that we can almost enter into a trance-like state called narrative transportation. And if anybody's familiar with the book by Daniel Counterman, his great book Thinking Fast and Slow, he talks about two systems, system one and system two. And really that system one is the emotion. It's our intuitive system that pre-processes information unconsciously before it's even past a system two, which is our analytical conscious mind. And so really what system one is always trying to do is trying to make sense of the world around us, is trying to make sense of the data and the information. And when we present our data and facts with a narrative, all of a sudden, it's easier for the system one to process and to make sense of.
Key Elements of a Data Story
Adel Nehme: Okay. That's so great. And I think this marks a great segue to really discuss the anatomy of a data story and the blueprint towards data storytelling that you describe in your book. Can you walk us through the key elements of a data story that you found and how they interplay between each other to create an effective data story?
Brent Dykes: Yeah, at a high level, there's a couple of versions that I have, but let me choose the high level version because really there's three elements that come together in a data story. And no surprise here. I mean, everybody's going to get this, but I think it's interesting to look at the interconnections between them. So the first one is data. Obviously, we can't tell data stories without data. And then the next one no surprise, visuals and narrative. Those three things. Now, the interesting thing is when we look at the intersections of these, we start to see the power of data storytelling. So we take data. If I give you a bunch of raw data, give it to a decision maker and ask them to make a decision or evaluate that data, there's a good chance they may not fully comprehend the data in front of them, they may interpret it in the wrong way, or they may get lost or come away very confused. And so what do we do?
Brent Dykes: If you think of it, it's like a Venn diagram where these bubbles, I have this diagram in my book, but when we overlap narrative with the data, what are we doing? We're helping to explain the numbers. We're adding in that extra context, we're adding in some explanation to help them interpret the data the right way and fully comprehend, the significance of what we're showing. Now, the next thing that we want to do, obviously going back to if we just had a data table, we dumped it on somebody's lap, there may be a good chance that they may not fully see the anomalies in the data. They may not see the patterns and the trends. And so that's why we visualize the data and that enlightens people to things in the data that they would miss otherwise, if it was just in a tabular format.
Brent Dykes: And then the last connection between these bubbles is, between the narrative and the visuals. And if you think about, probably, all of us are bingewatching some show on Netflix or whatever your local streaming service is, and that engages us, that combination of visual and narrative, that's why we watch movies, that's why we watch shows on TV. It's very compelling for us as human beings. Now, what I like to say about the power of data storytelling now is if we combine the right data with the right narrative and the right visuals, all of a sudden we have something that's very powerful, something super potent that can really drive change and can change people's perspectives. And so those are the three key elements that come together in a data story.
Weaving Narratives Into Data Insights
Adel Nehme: So obviously narrative is a key component of data storytelling. It's often the least emphasized when you talk about data visualization, data storytelling in general, and it's obviously one of the more difficult elements to master. What are some of the best practices you think data scientists can devolve when weaving narratives into their data insights?
Brent Dykes: Yeah, I totally agree that narrative is often the least emphasized and the most difficult to master I actually was on a webinar and we did a poll of about 100 analysts. And two thirds of the respondents indicated that it was of the three that I just mentioned, data narrative and visuals, two thirds of them agreed that narrative was the hardest to really work with and master. And so in my book, that was a key emphasis and I devote an entire chapter to looking at narrative and how to focus on a narrative structure. And I think there's lots of narrative models and frameworks out there that different people have interpreted different ways, but, but I basically break it down into four key steps. And the first step is to establish the setting, the context of whatever you're looking at, you might be looking at retail sales. And so you're showing what the trends are, what's typically expected. And then you have a hook as part of that first section where you're saying, but look what happened here. Sales went down or sales spiked up.
Brent Dykes: And so then it's like an observation that you're making in the data that will intrigue the audience and get them interested in, oh, what happened? Or what caused that? Or do we know what led to that? And so now you're hooking your audience, getting them interested in your analysis. Then you start to pull apart the issue or the observation to better understand it. Well, what could be causing the spike in this metric or this decrease in this metric? Unlike an onion, we're peeling the layers of the problem or the opportunity, and really diving into, I called this, the rising insights. We're starting to share a little bit more insights into the business that the audience can appreciate. Now, we're building up to what I call our aha moment. And you can think of that as the climax of your data story. It's your big takeaway. If the audience doesn't remember anything else, you want them to remember your aha moment.
Brent Dykes: And so that's your big conclusion. And typically, depending on the situation, it might be something that we've monetized. We might've said, here's a problem, and it's going to cost us $3.5 million if we don't fix it, or here's an opportunity and it looks like it's a half million dollar opportunity for this product line. Oh, wow. You get the attention of the audience. Now, the fourth and final step of this is the solution ain next steps. So we want to help the audience understand the urgency, but then also know what to do with it, and then that's where maybe some additional analysis is necessary to, hey, we've looked at the three options here. There's three options. Some have cons and pros, or maybe this is going to generate the most revenue for the least amount of costs.
Brent Dykes: We help position the audience to make a decision and feel fully empowered to do so. And so I think having a narrative framework like this can weed out information that isn't essential to your data story. One of the biggest problems that I think analytics professionals and data scientists have is sharing too much information, overloading the audience. And so having a narrative structure like this to guide you and say you know what is this really part of the flow of the story? Which of these areas doesn't make sense? Is it absolutely necessary or maybe no, it's supplemental information. If I got a random question from somebody, I'd want to have this available, so what do I do? I put it in the appendix. It doesn't need to go into the story. And I think that's one of the benefits that people don't realize about having a narrative structure or flow in mind when you're sharing your insights, is that it's going to streamline the information you're actually sharing.
Adel Nehme: That's great. And how strict do you think a data scientist should be when adhering towards a narrative structure or a framework? Do you think that this is something more of a plug and play and free around, depending on the problem at hand? Or is this something that is much more rigid to a certain extent when working with data and presenting data stories?
Brent Dykes: Yeah, I mean, I don't like to have a super rigid. My framework has training wheels. Once you've mastered that you can take off the training wheels because you know how to ride your bike, and I've had pushback from different people on my narrative model. Some people say, well, maybe I don't have the aha moment yet. Maybe I'm still in the rising insights where we're finding some information and we're targeting an aha moment in a particular area, but I'm like, yeah let's share that information. Let's share that with the business, get them on board. Maybe they redirect you in a different direction to a new aha moment. Another example is I've had some people push back and say, "I don't know what the next steps are. I don't know what the solution is. Maybe I would even feel uncomfortable proposing a solution to the business team."
Brent Dykes: Well, in that case, you share everything up to the aha moment and then you turn it into discussion and you get them to talk through things and then maybe you can then once they start to gravitate to a particular solution, then you can say, let me dig into that. Maybe I can do some additional analysis, verify, let's do some tests or let me gather some more information or data. Now that I have a clear picture what you think the potential solution is. So this is all just an idea. It is not hard and fast. And there's even the situation where I've had people say, "This is great. Having a story like this is great. It's not going to work for my executives," because they're not going to be patient to wait for a whole story to unfold. They're going to say, well, what's the number? What's the result. And so then what I did is, and I talk about this in the book, I talk about having not a data story, but a data trailer.
Brent Dykes: So it's like the worst movie trailer, because it gives away the climax. It gives away that so and so dies. But what you're doing with that data trailer, and really what we do is we just take the hook and we take the aha, we combine that into a shorter version of our presentation and what are we doing? We are asking permission for them to listen to the full story. And so we're giving them the soundbite, giving them the information, they can make a quick decision and say, oh, tell me more. Or how did you get to that conclusion? Or how do you know it's that? And oh, okay, well, let me walk you through the rest of the story, and now we have permission to take them into the story. So, yeah. That's kind of a three scenarios where maybe the model doesn't necessarily work, but we can modify, we can adapt. And I view it as training wheels.
Adel Nehme: That's great. And you've hinted at something here relating to audience like my executives require data trailer, rather data story. So audiences are also very important about what makes an effective data story. The priorities of a data science leader, for example, and the way they respond to information is not necessarily the same as a business account executive or a leader, and so on and so forth. How can data storytellers adjust their data stories to fit the audience's problems and their expectations?
Brent Dykes: Right. I mean, I always think that our data stories should be about the audience's problems. We should go into this and in my book, I talk about a simple framework that I like to use based on understanding four key dimensions of a particular audience. So if you know who you're going to be presenting to, or you know who you're going to be doing some analysis work for, there's four key areas you want to understand in depth and that's what is the problem they're trying to solve? There may be a top of mind problem that they're trying to address. We are losing customers, we're not retaining our customers or we're struggling to generate leads, for our sales team, whatever that problem. And maybe there's more than one problem, but really digging into understanding what is keeping them awake at night, the key stakeholders?
Brent Dykes: And then the flip side of that is, okay, so we know what the problem is, what is the desired outcome? Meaning, okay, so you're struggling to generate leads for the sales team. Well, what do you want to achieve? Oh, we want to double the number of leads that we're generating for our sales team this quarter. Oh, awesome. Okay. So now I know the magnitude of what you're targeting, what you're trying to achieve, that is super important. The third dimension is, okay, what activities, or actions, or initiatives do you have in place already to move you from this problem state to the desired future state? So in other words, if it's lead generation that's the issue, then they might say, well, we're completely revamping. We're really investing more in digital. Traditionally, we've done a lot of traditional marketing print, radio, media, and now we're switching to digital.
Brent Dykes: Oh, okay. So in terms of the actions and activities that are really important to them, in terms of their spending budget, they're looking at these things on a frequent basis, they're top of mind, they've got resources assigned, these are the areas that we want to investigate with our analysis, we want to probe and see, okay, let's see how their digital campaigns are working. And then lastly, the last thing, the fourth key dimension is the measures, the metrics, the KPIs. What are you really basing, like why is the problem a problem? Well, it's because we're underperforming on this metric. Or what's the target that you have for your outcome? Oh, it's based on this metric. Or how are you evaluating your different initiatives? Oh, we're measuring them by these, whether we're achieving these success with these metrics.
Brent Dykes: So I think if you could go into any situation to try and tell data stories, it's going to depend on your understanding of these four dimensions. And if you understand these really well, you're going to have meaningful data stories to share with this audience. You're not going to have to say, "Hey, you guys haven't been thinking of this, but here's something over in left field that nobody's thought of." I mean, they might say, "Well, our priorities are these and this isn't related to one of our priorities. So thank you for that interesting information, but we're going to get back to work on these priorities that we have." And so we need to be grounded in what's top of mind, what's important to them and then base our data stories on something that's aligned with those priorities.
Adel Nehme: So given that nailing a message that fits directly to the audience at hand is so important to the success and viability of any given data story, do you think that data scientists should work hand in hand in an agile and iterative manner with their audience on building a data story that fits their expectations and their needs? Or do you think a siloed approach is better off?
Brent Dykes: Yeah, I would say rather than working in a silo or a vacuum, I would always advocate to work hand in hand with the business, unless you have significant domain expertise and knowledge that, that business team has, it's going to be difficult for us to deliver impactful data stories without working with the business teams and engaging with them on some level. Obviously, you're not going to want to drag them through the weeds with you, but check ins and sharing progress and brainstorming problems or issues, I think that's great because at the end of the day, you're going to end up with a product or something that's, when I say product, I mean data story, you're going to end up with something that's going to be super relevant, super impactful, and right on target.
Brent Dykes: The worst thing we can do is go off on our own and spend hours and hours, come back with something that's not directly relevant to what they're focused on. And what does that do? That impacts our credibility, because you're not in tune. You don't feel the pain that we're feeling right now. We need help and you're not providing it. You're just doing whatever you want to do. So I think it's really important to be as aligned with the business teams as you can.
Organizational Data Literacy
Adel Nehme: I think going on top of that as well, it hurts the credibility of data science as a practice within any organization as well, when it doesn't solve the business problems or communicates the solution of data science. So one thing that's important as well that we discussed slightly earlier in our discussion that you mentioned clearly in your book, is how data storytelling is this ubiquitous skill that will define the future of work. So with that in mind, this is a skill that everyone needs to adopt, not just data scientists. How important do you view the role then of organizational data literacy when creating and consuming effective data stories?
Brent Dykes: Yeah, I think it's super important that organizations invest in establishing a basic level of data literacy, primarily so that you can have data storyteller. I mean, obviously data scientists are in a great position to tell data stories, but other people within the business, that are savvy with the data or at least have a knowledge of the data, can be put into a position to share insights and start to tell data stories. And I view data storytelling as a great way to cultivate the data literacy skills across the organization, because as more people are sharing and telling data stories, then more people are being exposed to data on a regular basis, and they're being helped with how to interpret it correctly. I view data storytellers acting as guides. They're walking people through the numbers and the more people that have these guides helping them to interpret the data the right way, it's going to only enhance and develop their own data skills in general.
Brent Dykes: So I see data storytelling going hand in hand with the literacy, both in terms of we need a basic level of data literacy to be a data storyteller. But then also as we have data storytellers telling data stories, that's going to foster an environment where the data literacy skills are just going to naturally be enhanced and helped.
Adel Nehme: I couldn't agree more. And I think this marks a great segue to discuss just how practitioners data scientists, or even business subject matter experts can become better data storytellers today. The most widely known element of data storytelling is something that we alluded to as well earlier in our chat is data visualization. What do you think are some of the best practices you can offer data storytellers when crafting data visualizations that help them anchor better data stories?
Brent Dykes: Yeah, I think the biggest thing that I point out to data professionals and data scientists is that, you need to recognize the difference between exploratory data visualizations and explanatory ones. And I've been saying for many years, that the chart that helped you discover an insight may not be the best one to communicate that same insight to others. And so you may need to either edit your visualization to make it more palatable to others, or completely shift it to a different visualization that perhaps communicates the data more effectively and clearly. You may have to remove some of the noise. When we're exploring the data, we're going to have all 10 categories, that we're looking at. But then when we isolate there's a problem with a particular category or two, do we need all 10? No. Maybe we just need the two that we found the problem in and maybe a couple others for context.
Brent Dykes: So those are the decisions. And I would say probably one of the most powerful tools that you have in your visual storytelling toolbox is color. And I would say, for example, when we're highlighting key points in our visualizations, we can use a bold color to really bring out a key data point to the foreground. We bring it to the foreground. And then what do we do with the rest of the data? Well, we might use gray scale to push it to the background. And so we can focus people's attention. Color is a super powerful tool for telling better data stories. And so that would be my one recommendation, definitely explore how you can leverage color more in your data visualizations, your explanatory data visualizations.
Adel Nehme: Couldn't agree more. I think really the power of color and the power of just using PowerPoint in presentation is so underrated than data science, but even across any profession that involves any form of public speaking. A key theme that cuts across as well in effective data storytelling is trust in data and credibility. I think oftentimes, when we talk about data storytelling as a persuasion tool, there is a risk of falling into the trap of finding data that supports a story or an agenda and not letting the data inform a story or an agenda. What are some of the pitfalls you think folks can run into when maintaining credibility of their data stories and how do you alleviate some of these pitfalls?
Brent Dykes: Yeah, I mean, being credible as a data storyteller is super important. Because if audiences don't trust us, it doesn't matter how good our data story is, it's going to fall apart because they're going to say, "Well, so and so, doesn't really validate their data very well." We remember that other story they told us where there's a bunch of errors in the data and we had a really bad... That can happen and we want to avoid that. Now, on the flip side of that, one of the things that I see is that sometimes on the analytics and data science side is, we're worried that we want to be perceived as credible. I would say we're almost too careful in the sense that we want to share with them. Okay, let me show you all the steps I took to reach this conclusion or find this insight.
Brent Dykes: And so we're walking... And I compare this analogy to, if we baked a cake, and we're like, okay, let me tell you how I sifted the flour two times before I... And I only used organic butter and all these little steps that we go through, nobody cares. At the end of the day, the audience typically is more focused on, "Hey, can I have a slice of that cake? And does it taste good? And hopefully it doesn't make me sick," but they don't care about all of these steps that you took to bake the cake. And I think the same thing goes into some of our analytics projects and data science projects is that we over-index on explaining all the details and trying to make sure that people are confident in our numbers and really, potentially what that can do is we can lose our audience, because they view a lot of this information as Charlie Brown's teacher [inaudible], and it's just noise to them.
Brent Dykes: Whereas, we want to make sure that we are always using integrity with how we approach our numbers and obviously being diligent on our checking things and making sure that we're double checking things, but I don't think that necessarily means that we have to bombard them with extraneous information that's not going to help them to understand the insight.
Adel Nehme: Okay. That's great. And obviously data literacy, statistical thinking, data visualization skills are really important when creating data stories. What do you think are the most important skills outside of those, that data scientist or even data practitioners of all kinds that need to learn to become effective data storytellers?
Brent Dykes: Yeah. I mean, all of what you've mentioned are really important. And then you also mentioned the importance of empathy, having empathy for our audiences, I think that's really important. I would also add critical thinking skills and curiosity, obviously, sometimes to find a good insight, you have to be very passionately curious as I've talked about. I mean, I think that was Albert Einstein mentioned that being passionately curious, he said, probably being super humble, but he said, I don't feel like I'm paraphrasing, but he says, I don't think I'm more intelligent than anybody else, but what I am, I am passionately curious. And I think that's a key skill for a data storyteller. And again, going back to the critical thinking skills, last fall, I taught a course on data storytelling, data visualization and I noticed there was a student that was really struggling in my class. And she was becoming increasingly frustrated with some of the real world assignments that I was giving them, because they were a little less defined.
Brent Dykes: And as I stepped back and looked at the difference between her approach and the other students that seemed to be performing well with the assignments, what was the critical thinking skills? I think she lacked really well-developed critical thinking skills to really help her excel on those assignments. And so I think that's also a key thing that we need as data storytellers.
Critical Thinking Skills
Adel Nehme: How do you address this lack of critical thinking skills in general? What are the best practices that you think could be integrated with an educational systems or within data science programs really instill these skills?
Brent Dykes: Yeah. I mean, I think one of the best courses I took in college was a critical thinking class, which was part of the philosophy program. And in I'll come clean on something. The first time I took it, I actually got one of my worst grades in my entire academic career. I got a C minus. And what happened was, at the time, it was my second semester and I was traveling to this university that was a long commute for me and I was doing it after work. And so I'd work full-time almost the whole week and then I would have a couple of night classes. And so this is a night class that I had, and I was so exhausted after working all day and then going to this class until eight or nine o'clock. I didn't hang around to do the office hours. I didn't ask questions of like, "Hey, I didn't totally understand this concept." And I just kind of like, I'm done, I'm just going back. And so I got a C minus.
Brent Dykes: And that was really frustrating for me. So I took the class again. Later, I wasn't working full time. I took the class, thoroughly, enjoyed it. Invested the time to really understand the logic and the reasoning. And for me, that was a phenomenal class that really opened my eyes to how, especially in today's world with the arguments, and how people form arguments and the logic and all that. It was just a fantastic course and I did get an A in it. So I went from a C minus to an A in that same class, so.
Adel Nehme: That's awesome, you redeemed that.
Brent Dykes: I redeemed myself.
Adel Nehme: You redeemed yourself. That's great to hear. So, another important aspect of storytelling is communication and presenting skills. What are some of the best practices there that you can offer practitioners to become better communicators in general? What are some of the skills or the ways you can improve that skill set?
Brent Dykes: Yeah. I think this is important. I mean, practice makes perfect. The more you do it, the more... And so maybe looking for opportunities to present your findings to people and seizing on those opportunities. And I think, I did recently a LinkedIn post about this and I said the one differentiator that I've seen that has the biggest impact is just practice. In some cases, a presentation is not going to require you to practice multiple times, it's not as critical. But if you're presenting externally at a conference or maybe it's a big executive meeting where a lot could ride on it, really going through your presentation over and over and over, to the point where you feel super comfortable with what to say on each slide, that's going to be huge for you, because then you can focus on other things like eye contact. You can focus on the movements of your hand gestures.
Brent Dykes: And sometimes you'll see when you're looking at conferences, you'll see people pacing a lot. It almost becomes an annoying pacing because they're not paying attention to themselves, because they're not realizing that they're nervously pacing. So when you've really got your presentation down, where you don't have to worry about it, you could do it cold. Then you can focus on some of these other things and you're just going to have a way better experience as a presenter and your audiences, when they see you confident in what you're talking about, it's going to make a huge difference for them too because they're going to believe in your numbers, they're going to believe in what you're presenting and they're going to have a higher, what's the word? Appreciation for who you are as a presenter and how committed you are to presenting the information effectively.
Trends in Data Storytelling
Adel Nehme: That's awesome. So before we wrap up Brent, what do you think are some of the trends that you see in data storytelling? What do you see the future of data storytelling and how it fits into the data scientist skillset?
Brent Dykes: Yeah, I mean, I see it as an evolving skill that still needs attention. Obviously, we're still working on data literacy, so a lot of organizations are still struggling with data literacy. But I see data storytelling coming shortly after that as organizations are starting to establish a certain level of data literacy, the next step will be okay. So everybody has a basic or most people have a basic understanding of data and we've democratized the data, most people have data in their hands now. How do we build a culture around, a culture where people are confident and capable of sharing compelling data stories. And so I see one trend will be more education on this area. Obviously, having a book in this area is great because I can help contribute to that education process that that companies and individuals are going through.
Brent Dykes: The other thing that I see, is how much is technology going to dip into the data storytelling world. We're starting to see that a little bit where you'll see some types of reporting are now automated with a certain level of descriptive texts and things. But I technologies that augment the abilities of storytellers to tell their stories is going to be really important, rather than full automation of data storytelling, where robots and algorithms are going to be telling all the stories. I don't see that necessarily anytime soon. And in a couple of areas where, algorithms and automated approaches are going to fail is having the full context of what's going on. Sometimes the data or the context is not reflected in the numbers, it lives outside of the numbers. And so, automating that will be challenging.
Brent Dykes: And then the other thing is tailoring to the audience. The presentation or the data story I tell to the IT team or the technical team that's going to be implementing a change, it's going to be much different than the story I tell to the C.E.O, or CXO, or even the middle manager in between those. So often we need to tell, even with the same insight's data, we need to customize that to the audience. So that's going to be another challenge for an automated approach, but I'm not going to say never. And I definitely think that, augmentation of what we're doing as data storytellers can be aided by technology, AI, all of that can be hugely beneficial.
Call to Action
Adel Nehme: That is very exciting Brent, thank you so much. Finally, Brent, do you have any final call to actions before we wrap up today's podcast?
Brent Dykes: Well, if you're interested in data storytelling, go check out my book, Effective Data Storytelling: How to Drive Change with Data, Narrative and Visuals. It should be available in a bookstore near you.
Adel Nehme: That's awesome. Thank you so much Brent. We'll make sure to include it in the show notes.
Brent Dykes: Thanks Adel. Thank you for the opportunity. It's been great.
Adel Nehme: That's it for today's episode of DataFramed. Thanks for being with us. I really enjoyed Brent's insights around effective data storytelling and all of the angles by which he can cover it. If you enjoyed this podcast, make sure to leave a review on iTunes. Our next episode will be with Rick Scavetta and Boyan Angelov and their new book by Python and R for Modern Data Scientists, and how it draws a path to the fall of the language wars. I hope it will be useful for you, and we hope to catch you next time on DataFramed.
← Back to blog