Latest / The Unpack Show with Anandh / Ep-14: Brent Dykes | Author of Data Storytelling
Transcript
- TheUnpackshow: Hey, welcome everyone for the at another episode. I have Brent with me. will let him introduce in a moment, but today's topic is close to my heart. It's about storytelling and it's more even close because it's about data storytelling. What we are going to do today is data storytelling is often described as a bridge between raw numbers and how we make real decisions. But today's episode is going to focus on what are the common pitfalls in data storytelling. So we are going to unpack that, why they happen, what they cause, how leaders and analysts communicate and communicators can avoid them. So that's going to be the topic for the day. Brent, I know you have a book that you wrote. You're an author of effective data storytelling. I'll give you an interesting story about that book. we, my team started a book club. So we meet on a monthly basis. The first book that we took was your book. Brent Dykes: Yeah. TheUnpackshow: It was interesting. we read like few chapters and come back on a monthly basis to review it. So it was interesting. So a lot of good nuggets that we got out of how to be effective in data storytelling. Thanks for that. And thanks for being part of the show. So why don't you introduce yourself and we can dive into the topic right away. Brent Dykes: Yeah, yeah. So Brent Dykes, I wrote, I'm the author of Effective Data Storytelling, and I've worked in analytics for two decades now. Most of my time, 12 of those years, were at Adobe, so the company that I was working for privately. Omniture was acquired by Adobe, and so that was 12 years there. Then I worked for four years at Domo, at BI, at Cloud. TheUnpackshow: Mm-hmm. Brent Dykes: based BI solution, and then briefly work for an agency before going out on my own and launching my own training, consulting, coaching business around data storytelling. TheUnpackshow: Right, right. Yeah, impressive. I think a lot of people think about data speaks for itself, but in reality, it rarely does. The moment you put some numbers on paper, we as data leaders or who is working in the data space will say, this should be pretty easy enough for you to understand. But a lot of times it missed the mark for the audience. Brent Dykes: Mm-hmm. Yeah. TheUnpackshow: Why do think so many data presentations are missing the mark in your view? Brent Dykes: I think it might be curse of knowledge. That might be a factor at play if you're not familiar with curse of knowledge for those that are watching or listening to the podcast. Curse of knowledge is really where it's hard for us to not know what we know. And so if we've spent a lot of time with the data, we've analyzed it for hours, days, weeks, we get very familiar with the numbers and what they mean in the context. we can, in our minds, we can distill the noise from the signal. And then we get excited and we share it with others thinking that they can process the information the same way that we do and see what we see. And that's often not the case. And so part of it is self-reflection and really putting ourselves in the shoes of the audience and thinking, how am I going to explain it to people who haven't spent it, know, haven't spent hours and days in the data like I have and maybe don't have the full context. So that's where the storytelling comes in. TheUnpackshow: Yeah. Right. Right. Brent Dykes: And that's why I think a lot of times data presentations can fail because we overlook that we know a lot more about the data than our audience does. TheUnpackshow: 100 % true. Numbers can overwhelm sometimes instead of enlightening the audience. Audience matters. But on the other end, you cannot just tell only the stories. You need to probably put the facts, is why the numbers matters too. But when we talk about the pitfalls, how does one strike a balance between the depth of analysis in terms of projecting the numbers and how they came up with the number versus clarity on telling the story? Brent Dykes: Yeah, think, well, one thing is obviously you have to understand what is meaningful and important to the audience. That's going to guide you in terms of what do you include in your presentation and what do you exclude. But I think that whole narrative arc or that narrative framework that you apply to your findings then helps you to kind of organize and really make those critical decisions like, do I include this or do I not include this? What goes into the story? What stays out of the story? If we have that narrative arc to kind of guide us, it's going to help us to really streamline what gets presented and what doesn't. What goes in the appendix? You might get a random question about this. Well, we want to be prepared for that, but does that have to be a part of our story? No, because that random question may not come up, and we would have wasted time anticipating and answering a question that nobody had. TheUnpackshow: Right. Brent Dykes: So I think it's really important to look at that narrative arc, that narrative framework, and decide, OK, this is what I'm going to share, and then obviously be prepared for those other questions. But I think that's a good thing. I really do think that the narrative component is not just about helping our audience to obviously engage with the content and have it be more memorable and more persuasive, but it actually helps us on the TheUnpackshow: Right. Brent Dykes: on the decisions that we need to make about what goes into the story and what stays out. And so I think there's a dual benefit. TheUnpackshow: 100%. Yeah, I think you said it really well, what goes into the story and what doesn't, right? That's the hardest thing to do. People always think about visuals are powerful, then where we age, right? There's always a debate about like, you know, how much we can put from a visual standpoint versus where we age. Sometimes charts are overwhelming too, but what do you think are some of the common mistakes that you have seen in the data visualization? Brent Dykes: Yeah. I'm trying to think what would be some of the comments. I think that there's this, I think a lot of people are like worried that they'll leave something out that'll be critical for the audience to understand, right? So often what I'll see when I'm looking, when I'm working with different clients and I'm reviewing their decks and kind of understanding how they share data today before a workshop or whatever, often what I'll see is these very dense slides, you know, where there's lots of text, there's maybe they're cramming in three visuals into the same slide, then there's just everything is, everything's labeled, it's just overwhelming. And so I think one of the key things is, one of the mistakes that's being made is just, it's almost like they're saying, I'm just going to throw all of this at the wall, and I hope something sticks, right? Kind of the spaghetti, we'll throw the spaghetti at the wall and hope that something sticks. TheUnpackshow: Right. Brent Dykes: But in doing that, in throwing all of that information at the wall, you're bombarding your audience with unnecessary noise that they just don't need. so it's really about, I kind of talk about it, you go through this exploration process where you're using visualizations and you're analyzing all kinds of data set, very robust data set, because you're trying to get that insight in the numbers. TheUnpackshow: Right. Brent Dykes: But then there's a transition. Once you've found an insight, once you've got something that you want to communicate to your audience, you then have to go through that process of editing out information that is too much, it's extra, it's extraneous, it's not necessary to understand the points you're making, and make that adjustment to really from exploratory to explanatory. And I think a lot of the time, people, analysts and data professionals, data scientists, TheUnpackshow: Mm-hmm. Brent Dykes: We just assume that what worked for us will work equally well for our audience. Again, that curse of knowledge coming in. But we do have to make an adjustment. We do need to remove the noise, maybe reduce the data set. Just focus on what are the key points and make those points as powerfully as we can. TheUnpackshow: Yeah, I think you said it well, exploratory to explanatory. In my view, context is the king, right? Context is everything. Probably like on the same lines, do you have an example that you can share for the audience where you have seen the business of the human context made a data story fail? The lack of business or the human context, yeah. Brent Dykes: Yeah. Yeah, I, yeah, I mean, I'll share a hypothetical, because I actually did a LinkedIn post about this today, where I just talked about, TheUnpackshow: Yeah, I follow, by the way, by the way, sorry to interrupt you, Brett. I follow your LinkedIn and there's a lot of good nuggets. So those who are listening, would strongly recommend to follow Brent because he shares a lot of good nuggets, you know, when it comes to data story telling. Sorry, go for it. Brent Dykes: Yeah, so one of the things I see is often, you know, people will see one data point and react to it. They'll jump to some judgment, right? They'll leap to a judgment or jump to a conclusion or whatever. And so the hypothetical example I share is that we, you know, I have a graph where I show there's six months of data and then the most recent, in the last month of data, the sales are down. And so I have an executive who's leaping to a judgment saying, it's marketing has dropped the ball again, here we go. And then the last action that they take is that they let some of their marketing team go because clearly they've dropped the ball in the most recent sales. And I call that the rush. That's a rush story. You're jumping to some judgment. if you work in data, in analytics, you're going to see this all the time. see somebody sees a statistic, they react to it. TheUnpackshow: Yeah. Yeah. Brent Dykes: They make some rash judgment or some rash decision. And that's not really being data driven. I mean, yeah, sure, a data influenced a decision, but not in the right way. And so then I contrast that with, well, with a data story, we're not leaping to judgment. What we're doing is we're walking with facts. in the bottom example, I take that same scenario where sales are down. TheUnpackshow: Yes. Brent Dykes: in the six months, but what do the leaders do? Well, they say, let's dig deeper. Let's go deeper. I want to understand what's going on here. Let's get that context to understand why the sales are down in the six months. And then when they start to unpack it, the first thing they realize is that the East region has had some massive storms that have interrupted their shipping for two months of that month. And so that's why the sales are down because they're they're not able to get the product into the stores. TheUnpackshow: Yeah. Brent Dykes: And then the second thing, when we look at the marketing team's efforts in the unaffected regions outside of this region that was affected by the storm, these other regions that weren't, we see that marketing actually had a 20 % lift on sales in those regions. And so rather than firing members of the marketing team, we should be celebrating that they were able to get a good result on their promotions and their campaigns. And it was this other factor that led to the lower sales was the storm, these severe storms in our East region. And so I think that just kind of points to how if we don't get enough context, if we don't dive into the numbers more deeply and really try and come up rather than jump into a conclusion, try and really understand what's going on and then make a more measured TheUnpackshow: Right, right. Brent Dykes: decision, based on context and data. TheUnpackshow: Yeah, I think you said it well when it comes to data-driven decisions. But I'm going to put you on spot, a little bit. So when you meet with stakeholders or when you meet with leaders, oftentimes you may see that they come with their own biases, right? Whether it's a gut-based bias or experience-based bias, you never know. But the real thing is going to be how you can elevate your data-driven story, right, in front of the stakeholders. How do you design data stories that both inform and also challenge this situation without alienating it? Brent Dykes: Yeah, I think it's important to really understand where the audience is in terms of, and actually written, I wrote a Forbes article on this, like how do we overcome that resistance? And so in that, I started with a four by four matrix, and then I was like, wait a second, there's actually a little bit of nuance here. So it ended up being a six, a two by six. Anyways, it's kind of, you'll have to see in the article, and I'm sure we can share a link to it. But in that article, I identified that there's different scenarios in which we react or our audience could react differently to the data that we're sharing. And in each case, we need to prepare for that. so that may mean that, for example, it's going to be very at a very simplistic level. It's a very different situation where the audience will welcome the results. They're actually excited. and they're going to receive the information positively. That is a much different scenario than if it was there, maybe they proposed some new program and they were super excited about it. And you have data that shows that that program was a terrible failure so far. Presenting that story is going to be very different, very different in how you handle that. And so a lot of it comes down to having empathy. TheUnpackshow: Ha ha. Brent Dykes: Knowing our audience as well. Like if we're going into that situation where we really don't know the audience and how they react. You know, we need to tread lightly. We need to be careful. It's much better if we if we know the audience, you know what? What are going to be the objections? You know, can we prepare for those objections? I mean, ultimately, you know why we can't control people's emotions and maybe we can't. Maybe they have a personal agenda or something that's not aligned with the company. and what the company should be doing. We have to navigate those waters, but I think it's much easier when we understand what we're facing. If we don't understand the position where the audience is coming from, we can be very reckless in how we communicate. And sometimes I wonder if there's analysts and data professionals out there that are communicating in a way where they really don't understand the audience and... and their ideas or their insights are coming across as facing some resistance and they're not doing a good job and how they adjust to kind of help people overcome some of those forms of resistance and get them to actually see it in a new light and maybe, and sometimes it's just the positioning, how we word, how we message. can make a huge difference. If we're blaming them, if we're putting them on the spot as this was a terrible decision that you made, how are people going to react? Whereas maybe we approach it differently. say, when we look at this, we rolled out this new program, we rolled out this new campaign or promotion, doesn't look like it worked for our customers. And you shifted away from placing blame to how do we serve our customers then? How do we fix this situation for our customers? And in that case, then we all want to contribute. Hopefully, some of those walls will come down, and they'll interpret and see the data differently. TheUnpackshow: Right. I think you talked about the trust aspect, the credibility aspect. But it is getting fragile in the age of data overload. But given the AI coming into picture, it is getting much worse. When you lose the trust, I think when you have the numbers wrong or when you are giving the wrong answers to the questions or the prompts, I think you will lose the trust right at that moment. But I'm curious to know, when you get into these kind of situations, right, because you're not going to be successful at the first run itself. So where do you see trust being lost most often in data narratives? And how can we, most importantly, how can we rebuild those? Brent Dykes: Yeah. Well, I think there's two. I kind of see data as the foundation of what we're doing with data storytelling. So there's two ways that you can erode that trust in the numbers. One is just data quality. If people are seeing poor data, there's inconsistencies in the numbers. Or maybe I've even been in situations where maybe the analysts made a mistake in how they're analyzing the data and introduced errors not Not because the data was poor quality, but because they did a poor quality job on the analysis or interpreting the numbers. So I think it's really critical to, obviously work with data that's going to be, that people will accept and trust. I that's probably, I would say, don't start telling data stories if nobody trusts the data that you're telling stories from. Like stop, do not pass go. Do not move on. If you have an issue with the data that you're telling stories from, that's never going to get better. You're never going to be able to make the data as pretty or engaging, because at the end of the day, the fundamental issue is that people just don't trust the numbers. So definitely, data quality is one. And then I would say the other thing that I think can happen, and again, it's that foundation, is that you are not presenting information that's relevant to the stakeholders, meaning TheUnpackshow: Right. Brent Dykes: If the data is not aligned with their priorities, what they're focused on, the objectives that they're tasked with, then you're just wasting their time. And that can lead to frustration as well. It's like, why are you telling me this story? I don't even, that's like the bottom of my list. I don't care about that. need to, here's what I'm focused on. I'm focused on these three things and you've wasted. 15 minutes of my time talking about this other issue that's not at the top of my list of priorities. so that's another area. If the data is not relevant, if the insight is not relevant, no matter how amazing we think that insight is, that doesn't matter if it's not relevant to your audience. so I think those are the two areas that you can make simple mistakes with the trust. Because it is, like what you said, it is a relationship that Hopefully over time as you work with these different stakeholders, as you work with different teams, and you're telling data stories, that over time you start to build that credibility. And you do that by having good, reliable data, relevant data, and then you're going to build a really strong relationship of trust. TheUnpackshow: Yeah, it takes time. It takes time. Yeah. Right, right. think it takes time, but once you have the trust and the credibility, I think you can just ride along the journey. But you need to make sure that you have to be consistent in how you tell the story. Know your audience, context is the king. Know what you need to put, what you should not put. A lot of good nuggets. So just on the fun side, I think some of the best storytellings or the data stories come from unexpected places. Have you seen anything? that landed surprisingly well and why? And what is the story? Brent Dykes: Yeah, mean, that was that's that's an interesting question. I mean, I think there's two parts to that. I mean, I always believe that. The really powerful data stories are built on an insight. And you might say, well, no, yeah, no big surprise there, but I I want to just clarify something and maybe to your listeners is that I think in analytics and in data science, we use the word insight very loosely, very almost too casually. And I've been trying to kind of beat this drum that we need to be very specific with how we use the word insight. part of going back to when I was writing my book, I was reviewing it with different people before I published it. two of the reviewers actually came back to me and said, hey, Brent, you use this word insight a lot in your chapters, but you haven't really defined it. And both of these independently said, you need to. define what you mean by insight. And I was like, this is great feedback. Awesome. And so then I went back and I was going to write something for the first chapter. And I was like, wait a second. What is an insight? I use the word very, very frequently. But have I really stopped and thought, what is an insight? And that's when I started to struggle. Because as I looked at different, I couldn't come up with my own definition. then I started looking at the dictionary definitions. And I was like, these aren't. very good. These don't, you know, they didn't ring true to me what an insight really was. And then somebody pointed me to a definition that was provided by an author and psychologist by the name of Gary Klein. And he defined an insight as an unexpected shift in the way we understand things. And for me, that's when my light bulb went off. That's when I had my aha moment. was like, oh, wow, that is, that is, that's a good definition. Very clean, very short, very concise. TheUnpackshow: Yeah. Brent Dykes: But it kind of represents really what I feel an insight is, is that the organization, the leader, myself, whoever it is, has some kind of understanding of the world. And then the data tells us it yields something different. kind of like unexpectedly kind of changes our perspective on something. So we see maybe it's our customers, maybe we see our competition, maybe we see the marketplace, maybe we see a business process that's broken or whatever. TheUnpackshow: Right. Brent Dykes: We see it in a new light. Our perspective has changed. so I think, going back to your original point, I believe that some of the best data stories are when you have an insight, one of those unexpected shifts in the, yeah. And then obviously the magnitude of that is, if it's something where the whole company believes that, TheUnpackshow: Yeah. Wow. moments. Brent Dykes: maybe this industry is, we can never compete in that industry, or maybe this particular customer profile is a big fan of our product. And then we've done some kind of analysis and we found, actually, it's not that customer segment, it's this other one related, but not how we currently define it. It's a smaller segment, a sub-segment of that segment that's really driving our sales. And then bringing that to the organization and sharing that. then I always talk about like, once you kind of have one of these unexpected shifts in our understanding, let's say with this particular segment, this more refined segment that we've identified, how does that cascade through the business? All of a sudden that changes how we market. No, we're not going to be marketing to this broader segment. We're going to be more precise and target this sub-segment of customers. And then we go to the engineering side. Well, we've been designing our product to service the needs of this larger segment. Well, what changes would we need to make to make it even more sticky and powerful for this sub-segment? And then how would that change our product roadmap? So there's this cascading effect. Once you have one of these really big insights, it can cascade through all aspects of the business from operations to finance to customer support to sales, marketing. mean, everything can be impacted. So that's really powerful stuff. But I think it all hinges on having an insight, a true insight, not just an observation as I kind of draw a contest between. TheUnpackshow: Thank you. Brent Dykes: Oh, we can see that sales went up 200 % last week. OK, well, that's the what. That's an observation. It's interesting. But until we understand the why of what caused that 200 % increase, we really don't have an insight. TheUnpackshow: Right. Yeah, think well said. Good way of looking at what is an insight, right? We often talk about an insight and action and a value. I think we have to ground ourselves on what does that really mean, right? Thanks for telling the story about your journey. for those who are listening, I would strongly recommend to read the Affective Data Storytelling book. Brent, I know you're a busy guy. Thanks for taking the time to be part of the show. Thank you so much. Brent Dykes: Thank you so much for having me.