Value Driven Data Science

[00:00:00] Dr Genevieve Hayes: Hello and welcome to Value Driven Data Science, where data professionals become strategic experts. I'm your host, Dr. Genevieve Hayes, and today I'm joined by Brent Dykes. Brent is the author of Effective Data Storytelling and the founder of Analytics Hero. He has consulted with some of the world's most recognized brands, including Microsoft, Sony, Nike, and Amazon, and is a regular contributor to Forbes.
[00:00:31] In this episode, you'll learn how to move beyond generating interesting findings to creating data insights that drive actual business change and how to use narrative as the vehicle to get there. Let's dive in. Brent, welcome to the show.
[00:00:48] Brent Dykes: Thanks for having me. It's great to be here.
[00:00:50] Dr Genevieve Hayes: One of the first requests I ever received as a data scientist went something like this, analyze this data set, and come back with any insights you find.
[00:01:00] To be perfectly honest, I didn't quite understand what it meant. Sure, I understood the analyze this data set part, but at the time I had no idea what my stakeholder meant. When she spoke of insights, I interpreted the word as a synonym for interesting facts, and that's exactly how she politely responded when I returned with my findings.
[00:01:23] That's interesting.
[00:01:26] But much to my disappointment, my work didn't drive any change. In fact, I think it just got shoved in a drawer never to be seen again as soon as I left the room. And in the years that followed, I found myself frequently reflecting back on that experience and wondering what I did wrong.
[00:01:48] But the answer didn't come until I eventually came across effective data storytelling. Brent, you argue that while most insights are interesting, not all of them create value, and that the ones that truly matter are the ones that offer some tangible promise of value, such as increased revenue, cost savings, reduced risk, and so on.
[00:02:13] While effective data storytelling is rightly celebrated for its data communication frameworks, for me, the most valuable part of the book was that very distinction. What it actually means for an insight to be meaningful or actionable, and how do we identify one, and this is exactly what I wanna explore today.
[00:02:32] So to begin with, Brent, what makes an insight actionable versus just interesting? Is it just that tangible promise of value or is there something more than that?
[00:02:43] Brent Dykes: My path to discovering what an insight is and what makes it. Actionable maybe it was a similar journey of discovery. When I was writing my book, effective Data Storytelling, I had some book reviewers that were reviewing the chapters and giving me feedback, and a couple of them both at the same time, said, Hey, Brent, you mentioned the word.
[00:03:03] Insight a lot in your chapters, but you haven't really defined it anywhere in your book. And so I was like, oh, that's great feedback. Thank you. That's perfect. I'll need to add a definition of what an insight is, early in the first chapter and then I went to go and write that definition and then I was like, wait a second, do I really know what an insight is and is it.
[00:03:21] Really clear in my mind and I struggled the moment. And then I started looking at the dictionary definitions and each time I read a new definition, I was like, oh, right, but it's not really resonating with me as deeply as it could. And so I kept looking. And then somebody eventually pointed me to a definition by Gary Klein, who's an author, I think he's a psychologist.
[00:03:42] And he wrote a book on insights called, seeing What Others Don't. And in that book he gives a definition that basically an insight is an unexpected shift in the way we understand something. And that was. Succinct enough and differentiating enough to really define for me what an insight is and what makes an insight special in the sense that it unexpectedly challenges how we view the world.
[00:04:09] We have assumptions, we have beliefs about. Whether it's our customers behave a certain way. Our customers like our product for these reasons. They buy it for these reasons. They're willing to pay up to this price, but no more. We have all of these institutional assumptions, organizational assumptions, even personal assumptions that go into something.
[00:04:29] And then when we find something that challenges, that the data shows something that is wait a second, how we view the world. Is not quite right. It actually may not be accurate and for me, that's the difference between. An insight and an interesting observation, an interesting observation is that maybe sales went up 120% last week.
[00:04:51] I don't know why maybe I haven't researched it enough or analyzed it enough to really find out why it went up, but that's interesting for sure. And why, I wanted to touch upon the insight first is I think we need to establish what is an actual insight before we even get into actionable.
[00:05:10] Because I think a lot of times we use that word insight very loosely, very casually about things like interesting observations or interesting data. Is there anything unusual? But there's a lot of unusual and interesting stuff that doesn't take us anywhere, doesn't drive any value, doesn't do anything for us. So I thought I'd answer and focus on what is an insight first before we get into the actionable insight.
[00:05:35] Dr Genevieve Hayes: I think a lot of this comes back to how data analytics and statistics are taught in universities because if I think back to the exams that I sat when I was an undergraduate, a lot of those involved. Here's a bunch of output. Graphs that have been produced by whatever point out what you see in them basically.
[00:05:55] And a lot of the answers to get full marks involve saying, oh look, this bar's really high, which indicates whatever. Or there's an increase in trend in this. So you get trained at university level without even realizing it. That a good answer is pointing out just. Facts in those graphs,
[00:06:17] Brent Dykes: Yeah.
[00:06:18] Dr Genevieve Hayes: there was never that connection to that's nice to know, but why would anyone care that line is bigger than this line?
[00:06:26] Brent Dykes: Yeah you're absolutely right. I think we are trained to find the anomalies, find the patterns, find the trends in the numbers and look at those. And sometimes, probably earlier in our careers that meant success. Hey, we found so. We found an anomaly.
[00:06:43] We found a pattern, a trend. And then the very next question that the executive or manager will say what's going on? Why is that happening? And then we're like oh yeah. It's not just good enough to point out there's something in the data, but go the next steps to actually investigate what's going on.
[00:06:59] And sometimes, as I talk about when I do my workshops and things, I say there's three outcomes of an analysis. One is you come back with an answer. Oh this happened because of x. And it's nothing shocking. It's nothing surprising. Doesn't change our assumptions.
[00:07:16] It probably aligns with our assumptions. Oh, it was because we ran that campaign that week. That's why those numbers were up. Oh, yeah. That makes perfect sense. It's an answer. It scratches the itch, so to speak, and I would say that's about 80% of the time are just in analytics and data.
[00:07:32] We're just providing answers from the data. How many orders were there last month? What coincided with that? Spike in number. Oh, it was 'cause of this. 15% of the time we might come back and say, I don't know what caused that. I don't know why that metric spiked.
[00:07:49] And at that point, it may mean because we don't have the proper context, it may mean that we're not collecting or have access to the right data. Maybe we need to collect more data. There's lots of different ways in which that will remain a mystery until we crack it. And in some cases I might actually go back to the stakeholder and say, do you want me to keep investigating or do you want me to prioritize something else?
[00:08:12] Because, I can dig deeper, but it's gonna take time and effort and They might be, oh, that's fine. Go over here, let's look over here. I wanna understand this better, or I wanna get an answer. So that's another 15%. And then the last 5% or less is those opportunities to really get an insight where there's something that shifts our understanding that's meaningful
[00:08:31] Dr Genevieve Hayes: I think you said that only around. 5% of observations are actually insights or less are insights. Going back to that example I gave with the stakeholder who was saying, get me insights. I want insights. She wasn't happy with. Just observations that I came up with from the data, but it's quite possible that there weren't actually any insights in that data set at all.
[00:08:54] How do you go back to someone and say here are some nice observations or. I didn't find any insights at all and retain some level of credibility.
[00:09:05] Brent Dykes: Sometimes I think it's about maybe explaining to that individual, here's what I did, here's what I looked at. We still haven't cracked the nut, so to speak. We haven't found anything. And I think maybe it's about educating the stakeholders this is a process this is a journey.
[00:09:23] It's like fishing, you're not guaranteed to always hook a fish. You can go fishing and you can come back from that trip empty handed, and then other times you get lucky, hey, I got a fish and then I got another fish and I got another fish.
[00:09:36] Maybe I'm using the right bait. Maybe I've got the right location, maybe the right time of the day. There's all these different factors that go into making a fishing trip successful. And using that metaphor not all of the times that we are analyzing the data, are we gonna have.
[00:09:54] An answer and be able to explain everything that we're seeing in the data. And I think maybe immature managers who don't understand what it takes to do that analysis and aren't patient maybe will become frustrated with that process. But I think it's about mentoring, coaching, maybe sharing what I have.
[00:10:15] I wasn't able to determine this, but I was able to determine these three things. And so coming back with something and maybe that's I don't know why that happened but here are the things I've ruled out. It wasn't caused by this, or I looked over here and that wasn't a factor, and so you can start eliminating things. So I don't think it's coming back with I don't have an answer. It might be a middle ground where you're saying, here's what I know and here's what I still don't know.
[00:10:41] Dr Genevieve Hayes: I think the thing. It's key here is you are treating the analysis as identifying answers to questions. So this is aligned with that GPS analogy you have in your book where you've got a focus for. Your fishing expedition to begin with. So you're not just looking for random things.
[00:11:04] If I look back at that piece of analysis I did early in my career, I think that was my problem. I was just working through all the different. Data visualization techniques that I'd been taught in my studies and I didn't have a focus for my analysis. I think if you have that focus, then you've got a much higher probability of coming back with something useful.
[00:11:25] Brent Dykes: Yeah. And that's something that I feel is really important just in my career in working analytics and working as a consultant. I talk about the four. D framework in my book the four Ds, for those of you that haven't read the book there's four key dimensions and these are applicable to any project, whether it's a dashboard project, an analysis, a project, a reporting project, a data storytelling project.
[00:11:47] If I can understand these four key dimensions. I am so much further ahead in being able to deliver value to the audience that I'm targeting or the stakeholder that I'm working with. The first one is problem, so the better I understand the problem that they're trying to solve, the problem that's keeping them awake at night the more I understand that problem, the better analysis I can do and the more focused I can be in terms of where I look in the data.
[00:12:13] The next dimension is the outcome. What outcome are they trying to achieve? Maybe they have a problem with generating leads for their business. Maybe it's marketing department. And so that's the problem. And then they have a target. They have some kind of goal, where they're trying to get double the number of leads that they're generating for sales this quarter.
[00:12:32] And so at that point that's helpful as well. 'cause then I can say, oh, I can see how much progress they've already made. What's the gap? Is it, oh you're almost there. It's a very minor tweak to get to that doubling of your leads or you look at it and you say, oh my gosh, you're, there's no way, you're really far off your target.
[00:12:49] So there's a lot of work that needs to be done. So that's the second dimension. The other third dimension are the actions or the activities of where are they spending their time, where they're making their investments in resources, budget, that's gonna guide where I'm gonna focus my efforts because I can look at, okay, from a marketing perspective, you're doing a lot of online events, you're doing a lot of let's make something up in conference events.
[00:13:16] And so those are where you're placing your bets, and there may be ways in which we can enhance those efforts and there may be problems with those efforts that we need to investigate. So rather than looking at. All of your marketing efforts and maybe even marketing efforts you're not doing we focus on where are you placing your bets today and let's evaluate how successful or unsuccessful those are.
[00:13:40] And then the fourth dimension. Is the metrics or measures. So what are you judging the performance by in terms of the activities? How are you analyzing the problem? What metrics are you using for that? Maybe there's targets that you're associating with. And so again, the measures come into affect the outputs.
[00:13:58] But if I can get a holistic understanding of all four of these dimensions, I now go into the data with a much better, clearer purpose. Direction. I can dismiss data that's not related to, what the objectives or priorities are of this stakeholder based on the problems and outcomes they're trying to solve or achieve and.
[00:14:22] That just gives me way more focus as I'm doing analysis and as I'm working with the data and it increases my chances of coming up with an actionable insight because going back to your very first point. I'm focused at this point. I'm going in with a mission, and hopefully I'm coming out with something that could be valuable to the audience.
[00:14:42] Dr Genevieve Hayes: And that brings up the topic of actionable insights. So I don't think we've actually covered what is an actionable insight.
[00:14:49] Brent Dykes: So when I started thinking about actionable insights I really liked the work of a guy by the name of Ash Kic. And he was an evangelist at Google for a long time, and he came up with a framework called Care Do Impact, and he called it the Care Do Impact Framework.
[00:15:06] And when he was managing his analytics or data science team, he said there was three things that we need to think about when we have an insight, if it's actionable. One is, why should your audience care? That's the first point. Second, what should they do about it? And then third, what's the potential business impact?
[00:15:25] And what I did in my book and what I've done in my consulting is I've taken those three questions that i've basically found two criteria for each of those questions. So for that first question, why should your audience care? And the two criteria I associate with that is it's gotta be valuable.
[00:15:41] And it's gotta be relevant, so if you're focusing on something that's valuable to your audience, meaning it's aligned with their strategic goals, it's relevant you're gonna get their attention. And so if it meets those two criteria, you're on the right path. Related to that second question, what should they do about it?
[00:15:59] In order for your insight to be actionable, it has to have a solution that's practical and specific, I could have an insight and it could mean that if we invested $2 trillion in some new technology we'd make a bunch of money.
[00:16:14] But it's very unlikely and impractical to get an investment of $2 trillion. And so it's thinking about is there a practical solution? And how specific can I be if I'm ambiguous or vague about how we're gonna implement this insight again. That to me, erodes the actionability of an insight.
[00:16:36] And then the last two criteria tie into the third question, which was what's the potential business impact? And I think sometimes on the data side, we overlook that it is self-evident to us that something is valuable. Something is gonna have a good.
[00:16:53] For the business, but we haven't put it in concrete terms that the decision maker can understand and we haven't contextualized it. And so in terms of making something concrete, I always go back to can we monetize? Can we quantify the impact of making this change? It could save us $2 million over the next year.
[00:17:15] It could generate a half million dollars over the next quarter. Whatever that figure is, whatever that calculation, even if it's an estimate, a potential. Sometimes they get pushback from some data people. It's I can't guarantee that. I don't know for sure. And so they put nothing.
[00:17:31] They put no quantification on it, and I think that's probably one of the worst mistakes you can do because it's not real to people until they can think about it in terms of. Money, dollars and cents what this is gonna mean to us. You can couch it all you want as potentially, there's a chance you can even give a confidence range of it could be between 1 million and $3 million, but even on the conservative side, you probably still get somebody to say, oh, a million dollars.
[00:17:58] We should definitely explore that. And then the other thing I say contextualize because sometimes we put out a number and we don't contextualize it, meaning, there's lots of different ways you can contextualize an insight but saying things like, that's more than what we spent on marketing all of last year, and one insight can pay for itself and pay for all of the marketing we did last year just from that one.
[00:18:23] Action or in my training I talk about eight different ways in which you can add context. But context really helps the audience to understand the significance of the number and the urgency of the number. And so I feel like the concrete in the contextualized go together.
[00:18:38] So if you have an insight that meets all six of the criteria, it's valuable, it's relevant. The solution is practical and specific, and you can put it in concrete terms and contextualize it. You have an actionable insight that's has a good chance of driving some kind of change in an organization.
[00:18:57] Dr Genevieve Hayes: A key message of your book is that effective communication is what turns insights into action. And this is done by effectively wrapping a narrative around an insight why narrative and not say just visualization or a well-written executive summary.
[00:19:14] Brent Dykes: Yeah, I get that a lot obviously a lot of executives expect executive summaries and I haven't written this article yet, but I will write it this year. 'cause I've already shared it in my workshops and things, but there's a difference between an executive summary and a story.
[00:19:31] Okay. So a lot of the times you might've heard of the pyramid principle the McKinsey model of sharing information. So you put your conclusions up front and then you back it up with supporting evidence. And so it's very conducive for executives because you're getting. The key message in matter of minutes.
[00:19:49] But there's a trade off that nobody talks about when you're solely relying on executive summaries and I think the trade off is between efficiency and effectiveness, so it's very efficient to summarize everything as an executive summary. And it's something that I talk about in my book.
[00:20:06] It comes up so often that I have to address it where some people say I don't have time to tell a data story. They're not gonna give me an hour or 30 minutes, or even 20 minutes to go through this. I have to tell them in five minutes or 10 minutes what's going on. And what I talk about is creating a data trailer where there are key moments in your story that you're highlighting and you're leaving others out.
[00:20:29] But the goal of that is. For that to peak their curiosity, to peak their interest in hearing the rest of the story. And so that's different than an executive summary. So an executive summary the efficiency is you get all the numbers, you get the conclusion upfront with a data story. It's about effectiveness.
[00:20:48] What are we gaining when we tell a data story? We're deepening their understanding of the topic, we're actually walking them through the, in a cohesive and salient manner. The key points of our. Discovery. Like it started here, we found that this metric was going up or down and it dropped or spiked dramatically.
[00:21:10] And then we looked into that and we found this, and then that led to this and this. And then we have our aha moment, which is our big insight that we're sharing. Now the difference between that delivery and an executive summary is that the executives are really gonna understand the problem or the challenge or the opportunity.
[00:21:30] 'cause we're able to walk them through it at a deeper level. And so the depth of understanding also the. Ability to basically connect with them and have them emotionally understand what that story would mean to the business and to them is much richer and deeper. And so it's gonna be far more persuasive than just a summary, which.
[00:21:53] Is efficient but hasn't really walked them through all of the details that the depth of understanding will be much shallower and honestly, in some cases there may be situations where because they don't fully understand the challenge or the opportunity like they would from a data story.
[00:22:10] There's a chance that they make the wrong decision or they overlook something or come away with it with the decision that maybe it is based on an incomplete. Or insufficient understanding of what's happening. And so they're missing the context, they're missing the connective tissue that ties it all together and explains why, we're looking at a $3 million opportunity if we fix, X, Y, Z.
[00:22:36] Dr Genevieve Hayes: I've worked in environments where a lot of decisions are driven by board papers or similar committee papers. So would the best strategy in those sorts of environments be. Write the paper using the standard format, but when you present the paper, use that data storytelling approach that you advocate for.
[00:22:59] Brent Dykes: I and just to give people an overview of what that approach would be, it's not oh, we're gonna tell 'em, a movie. Like narrative? No it's, we're taking elements from the narrative framework, a narrative arc. And so like a movie or a novel, you have to start with establishing the setting,
[00:23:16] and so we go into establish what is the context of what we're looking at here. Very briefly, again, one of the mistakes a lot of people will do is they'll start to tell what I call the analyst journey, and they'll start to talk about how they did their analysis and what they looked at. No.
[00:23:31] That's not what I'm talking about. That is definitely not the setup. It's a setup from a business perspective. How does this relate to their goals, to their objectives? What's the context? What's the slice of the data we're looking at here, these are all of our return customers, over the last three months we've noticed a trend where something's happening and so that's establishing the setting.
[00:23:53] And then we go into what I call the hook. And the hook is really just a major observation. It really is an interesting data point that we saw, in some cases that even argues sometimes the stakeholder might have been the one that gave us, they looked at a dashboard and they said I don't understand why this is trending up.
[00:24:10] For the last three weeks it's been trending up. That doesn't make any sense. That's weird. Could you look into it? That's my hook. I'll come back with it. Four weeks ago you asked me about why this is trending up over the last three weeks. Here's what's going on.
[00:24:22] And then from there, once we have our hook, whether they provided it or we're the ones providing it we now hopefully have something that interests them in hearing the rest of the story. They're like, oh. I'm curious. I wanna know what caused that spike or that trend to trend up over the last three weeks.
[00:24:37] And then for there we go into what I call the rising points I changed the name after publishing the book, but I'll just use the name I use now, which is Rising Points. Think of it like an onion, where we start to peel back the layers of
[00:24:48] the onion to uncover what's going on. We're connecting the dots for the audience. And then that again, doesn't mean that we're showing them all of our analysis. No. We're just picking out those key parts that connect together to tell a story. And then we build up to our aha moment, which is our main insight and we're not done with the insight.
[00:25:09] The aha moment also includes the so what? So the reason for them to care. The quantification of, what this means, if we make a change or we do something. And then the last part of the story is the solution and recommendations or next steps. So what do we do about this?
[00:25:25] And often that could mean additional analysis of there's three options here. We could go very conservative, we could go very aggressive, or we could go somewhere in the between or maybe there's three options here, whatever. And we review the cons and pros of those options and then hopefully.
[00:25:41] Through this whole entire process. My vision of data storytelling is it does two things for the audience. One is they're basically learning about the business. Whether that's, customer behaviors, processes within the business, competitive actions or intelligence that are meaningful to the business.
[00:25:58] They're learning. They're learning something that they didn't know. And then two, because of. The way we've packed this up by including options, by including recommendations, by sharing all of the information context they need to make a decision. We put them in a place where they can make a decision.
[00:26:16] Not just oh, let me go back and think about it. Hopefully if we've done a good job, we've given them everything they need to start going down that path of making a decision and taking action. I've worked in analytics long enough to know we collect a lot of data, we visualize a lot of data.
[00:26:33] It goes into dashboards, goes into reports, and often. Nothing happens. Nobody does anything. And that could be for a number of reasons. We're not collecting information they care about. It could be because the information's too overwhelming. It could be because they don't know what to do with it.
[00:26:50] And so that's why I see the place for a data story, especially when we have an insight. I always emphasize that in my trainings is that at the crux of a data story, it's not just about sharing some information. No. We have a real insight and if we think about an insight and if it's actionable, it can have a meaningful.
[00:27:10] Impact on the business. It can change and affect the business or the organization positively if it's implemented. And so I think understanding the narrative and how that would work, how we communicate the insight using a narrative arc can really make a meaningful difference.
[00:27:29] It can make the difference between throwing insights over the wall and hoping that somebody takes an action and actually playing a role and seeing that insight drive change. And the subtitle of my book is, how to Drive Change.
[00:27:43] With data narrative and visuals. That's the role we need to be playing. And if we are, then we become a strategic advisor for the business that is gonna pay dividends for both our careers and for the organizations that we work for.
[00:27:58] Dr Genevieve Hayes: So what is the single most important thing a data professional can do starting today to become better at generating and communicating actionable insights?
[00:28:07] Brent Dykes: I think if there's any gap between your understanding of what's important or valuable to your audience 'cause I run into that all the time. I come from a business background,
[00:28:17] I don't come from a STEM background. I come from a marketing background. I started out my career as a marketer, and then I went into analytics, and I always felt like I had quantitative skills, but I didn't have a quantitative degree. I had a marketing business degree.
[00:28:33] And often I'm working with technical people who have amazing analytical skills, technical skills, but they don't understand the business, they don't understand the numbers, the metrics that they're working with. They don't understand the priorities, the strategic.
[00:28:49] Initiatives that are going on within their organizations. And I think for me, if I were to recommend, one area of focus would be to try and learn the business as much as possible, or learn what's important to your audience and start to invest in that area because that impacts everything.
[00:29:06] Even as you're analyzing, if you understand a little bit more about the business or even I would say if you say that's impossible, Brent, I can't learn the business. Then partner with somebody on the business side. There may be a marketing manager, there may be a HR manager, somebody who can be, that GPS that you need for the business side of things and you partner with them obviously you can handle the analysis and the collecting of the data, the analysis of the data, but you have them on your shoulder guiding you through those numbers and really focusing your attention on what matters and what doesn't matter.
[00:29:41] Dr Genevieve Hayes: And for listeners who wanna get in contact with you, Brent, what can they do?
[00:29:45] Brent Dykes: I'm on LinkedIn, so you can follow me there. I'm usually publishing new content three times a week. And we'll see if that continues. It's getting tiring, I've almost been doing that for five years now. I.
[00:29:57] But that's a great way to find me or go to effective data storytelling.com and you can find me there. There I do some blog posts and I have all my services and my link to my masterclass course on data storytelling there.
[00:30:11] I'm always working with organizations, helping them and individuals. To get, these skills into more organizations so they can hopefully capitalize on all the rich and wonderful data they're collecting and hopefully get those insights communicated effectively so they can drive the change that they need to.
[00:30:28] Dr Genevieve Hayes: And that's it for today's episode of Value Driven Data Science. But if you want more from Brent next week, you can catch our Value Boost episode where we explore the impact of AI on generating insights. And if you haven't already read Effective Data Storytelling, I can't recommend it highly enough.
[00:30:46] Thanks for joining us today, Brent.
[00:30:49] Brent Dykes: Me, it was great to be here.
[00:30:51] Dr Genevieve Hayes: And for those in the audience, thanks for listening. I'm Dr. Genevieve Hayes, and this has been Value-Driven Data Science.