Speaker 1: This is Catalog & Cocktails. Presented by data.world.
Tim Gasper: Hello everyone. It's Wednesday once again and it's time for Catalog & Cocktails presented by data.world. It's an honest, no BS, non- salesy conversation about enterprise data management with tasty beverages in hand. I'm Tim Gasper, longtime data nerd, product guy, customer guy, data.world, joined by Juan.
Juan Sequeda: Hey Tim, I'm Juan Sequeda, the principal scientist here at data.world and as always Wednesday, middle of the week, end of the day and time to chat about data. And today we're going to have a conversation which I've been wanting to go have for so long, which is about games and sports and data and books and education. I'm super excited to have our guest, Maddy Want, who's the VP of data at Fanatics Betting& Gaming and author of the recently published book Precisely Working with Precision Systems in a World of Data. Maddy, how are you doing?
Tim Gasper: Welcome.
Maddy Want: Doing well, thank you both for giving me an excuse to drink vodka on a Wednesday. Happy to be here.
Juan Sequeda: Well you could always listen to us live and always have that excuse to inaudible to do that.
Tim Gasper: And you're making up for us because we're in the office right now and unfortunately our cocktail game here is not so good. We got some beers. So Maddy, thank you for upholding the ideals of this podcast.
Juan Sequeda: So we'll start with that. What are you drinking and what are you going to toast for today?
Maddy Want: This is a vodka grapefruit situation. It's very summary, very refreshing and I'm going to toast to... Well, if I can make it serious for a second. May 17th is the International Day of the Child helpline, which is a cause that I think everybody on planet earth would agree is valuable. It's intended to raise awareness about all of the helplines and hotlines that are available in various different countries for children who are in various states of need or distress to call. And I think it's awesome. The work they do is awesome and it's worth a cheers. So that is what I'm cheering to.
Tim Gasper: Definitely cheers to that.
Juan Sequeda: I'm learning something new. It's May 17th. Cheers to that.
Tim Gasper: Cheers.
Juan Sequeda: We're having a beer, I'm having Cerveza Set brewed with lime from Firestone Walker. I think it's all kind of a Texas one. It's actually not that bad.
Tim Gasper: Yeah, kind of light.
Juan Sequeda: I'm surprised. I'm surprised. I'm enjoying it. How about you?
Tim Gasper: I'm drinking a Revolver Brewing Blood and Honey, which is another Texas beer. And this one is a little bit on the sweet side, so it's kind of a light sweet lager.
Juan Sequeda: So in addition to toasting for May 17th, the Child Help Line it, we're continue to celebrate. It's been three years on Sunday officially it was three years, 130 episodes. I think over a hundred guests of doing Catalog & Cocktails. So just thanks to everybody that's been listening, it is amazing. I'm super excited. We're both going to be next week at Gartner, so catch us at Gartner in London, so that'll be a lot of fun. But hey, let's go to our warmup question. We're talking about sports today. So what is your favorite team mascot in all of sports?
Maddy Want: So I had to look this one up to make sure I got the name right, but the first game that I went to when I joined this job was an NHL game with the San Jose Sharks against somebody who I can't remember because they lost and it didn't matter. And I don't know if you know this, but the way that the Sharks come out onto the ice is through this giant sharks head and they have a shark mascot whose name is SJ Sharky, SJ standing for San Jose and he's awesome. And so ever since then, because that was my first game in this job, that's got to be my favorite mascot.
Juan Sequeda: That's a good one. I'm actually originally from San Jose.
Maddy Want: So you know Sharky.
Juan Sequeda: I'm familiar with Sharky. Yeah, as a kid, go season games. How about you Tim? I guess I think I know what you're going to go say.
Tim Gasper: Well, I grew up in Cleveland, so for me, I have a special connection to some of the Cleveland mascots and I used to go to what were called the Cleveland Indians games now the Cleveland Guardians. And the Cleveland guardians had Slider, which was their mascot. And I always enjoyed seeing Slider throw the T- shirts out into the stands and things like that. So that has a special memory for me.
Juan Sequeda: And I am a UT, University of Texas grad, I'm a Longhorns, we have Bevo. I think just being at the stadium and actually having a longhorn there, that is something that it's just always so exciting. And it's always exciting just going around, you're traveling around the world and you carry the Longhorn, people will see that and they'll say, " Hey, hook them." That's really cool. So we're very supporty today.
Tim Gasper: I know we all have our sports connections here, which is perfect.
Juan Sequeda: All right, well let's dive in. Maddy honest, no BS, what are your favorite stories organizations have been criminally under leveraging their data.
Maddy Want: Yeah, I did say that, didn't?
Tim Gasper: We didn't mean to criminally under leverage your data?
Maddy Want: I think it's going to sound obvious. Anybody who's listening to this podcast just sort of self selectively is going to be somebody who probably would agree with this statement. But any organization that's not treating the data as an asset is criminally under leveraging it. That's sort of categorically true. I think, I hope that as a digitized world we're moving past overall. Anybody not realizing yet what data can do for their organization, but sometimes I still get surprised. And I've spent the last four years or so writing and publishing a book about just incredible uses of data across a range of different industries, public, private, nonprofit sector. And just really, really creative and transformative. And I wish everybody could know these stories and just be imaginative about ways that they can leverage their data. It's not just all about reporting and business insights, it's about thinking of it as something that has a dollar value and being able to find ways to make that come true.
Juan Sequeda: So let's unpack a couple of things here. And this may be a basic thing, but let's define it. What do you define treating data like an asset? Because I know when people talk about treating data as a product and all that stuff, what is your definition here about this? Make sure we are all on the same page.
Maddy Want: It's closest to the accounting definition. Which side of the balance sheet are you listing it on? I started in tech 13 years ago now, and at that point data, or at least in the companies that I was working at at the time, data was a cost center. Everything to do with data was treated as a cost center. And that's because data wasn't making sales, data wasn't the product that people were signing up to. Data was a secondary or tertiary function in a business who so often wasn't actually able to do what they needed to do and what they needed to do was undervalued anyway. And so I feel very lucky to have continued to work in tech over the last decade or so where that has basically completely changed. And now data is leading strategy at so many organizations. On the other side, on the veering two side or the other of the spectrum, there's an interesting conversation to be had about whether there is any intrinsic value in being data driven in and of itself. Like, " Is that a goal? Why would that be a goal? Is data sort of an end or is it just a means to some other end?" And I think sometimes the sort of adherence to being data driven can also be problematic.
Tim Gasper: That's interesting. I know a lot of folks still use that language today where they'll say, " Our goal is to be data driven." Or, " Our goal is to increase our data literacy." Do you see that as an end or as you've gone through your experiences, do you see that as that's a part of the journey and there are other factors here?
Maddy Want: It's more like a must do. It's more like the goal is to... If you're in the B2C tech industry like I am, the goal is to delight customers. The goal is to build something that's of value to people in the real world. And you got to know your way around data in order to make that happen. But that's not the end goal.
Juan Sequeda: So this is super important, understand how the business works and you're in your business, a goal is to delight customers. So whatever you're doing with data, it better freaking delight customers, otherwise it's not providing value, it's not an asset. Then if it's just reporting and BI reporting and stuff like that, then it ends up being that cost center. And I would argue that we're still not there yet. I would argue that the majority of any organization, they think about being data driven when really it's like you want to be able to use data to generate new value. But by saying they're being data driven, they're really just doing the BI reporting and then focusing on the infrastructure to make sure that this data that generates this dashboard isn't broken or whatever. But it's like that's not delighting your customer. I mean it's probably helping your operations folks that eventually go do that, but it's a very indirect way to go delight your customers, help the organization make more money or save more money. And I think that's the disconnect that we need to go have. And I still think that we're in two groups. You are definitely kind of in the group that is delighting customers while other people are like, " It's just part of the operations." I don't know, I started to rant, I'll shut up.
Maddy Want: Everybody who shops on Amazon, which is clearly one of the most data- driven companies in the world, it has this experience of you go on to buy a toaster and so you search through a bunch of toasters and you leave and you get retargeted with a bunch of Amazon toaster ads. And then you come back and you buy one of the toasters and you continue to be retargeted with Amazon toaster ads. And there's a whole bunch of reasons why that happens. And there's a lot of smart people. It's not like this is easy to solve, but it's just one of the most crispy ways to help people who are not from the data industry understand what we mean when we say delighting customers is the point of data. Because it is not delightful to be followed around for an ad for something you've already bought. Especially when you feel the power of Amazon's personalization capabilities in other ways that are so striking. And then you have this one particular retargeting experience and you think, " Gosh, okay, this is not ideal."
Tim Gasper: In order to have that perspective, you actually have to step out of the... you have to see the forest from the trees. The tree that you see in front of you is, " Oh well optimal thing here is let me show the toaster after you look at toasters." Because maybe when you look at a small enough range of situations, maybe that does result in the most clicks or something like that. But when you truly zoom out and you look at the bigger picture and you think from the outside in perspective, what does the customer feel right? And what does our business actually care about? Does it only care about this transaction right now or does it care about the customer lifetime value? Then all of a sudden your perspective changes. But that requires perspective. You have to understand the business and you have to think beyond just the now.
Juan Sequeda: Yeah. There's so many stories in your book, but before we get into stories, I just want you to share your favorite stories. Give us some background about the book because it's very, very impressive and I mean it's a huge motivation.
Tim Gasper: And how long have you worked on this book? A lot of efforts gone in, right?
Maddy Want: Four years end to end, four years to almost the day, but definitely to the month. It was a bigger endeavor than I could have ever imagined. But you get into it and we, myself and my co- authors, Zach Tumin, from an academic background and he's also from a variety of different industry experiences. And one of the most recent and relevant was that he was the deputy commissioner of the NYPD. And through that association, the experiences that he's had and also the way that I met Zach was because I was studying at Columbia SIPA at the time. And he was my professor there, opened up this whole world of public sector and nonprofit data stories that I think if left to my own devices I wouldn't have thought to go there. I wouldn't have focused on that as much because my career has been in private industry the whole time. But some of my favorite stories came out of the public in the nonprofit sector and I'd love to tell a couple of them if we get time. But the process of writing the book was a year of thinking about the concept and breaking it down and really making an argument for what we were going to do and why, selling that concept to someone. And then two years of interviewing fantastically interesting people about what they've done. And then a year of writing that up and editing it and editing it and then editing it a couple more times. And then here we are finally four years later it's getting published. But it was an experience. I got to say, I might not be on the eight, nine or 10 on the NPS scale of what I recommend writing a book. I might be a detractors on that scale.
Tim Gasper: Well I'm glad you have because I think there's a ton of incredible stories that you've collected through this experience and maybe it makes sense to jump into some of that. Maddy, when you think about your book and what you've been able to surface, what maybe let's start with your favorite story that's kind of come out of all of this. What's your favorite story that you ran into and were able to bring to life here?
Maddy Want: Well, I don't have any favorites, but I have two favorites and one of them is a story about an incredible company called Zipline, who I think actually in the years since we started writing the book has become more well known, which is good. They deserve it. What they've been doing is pioneering drone delivery of blood and medical supplies to remote parts of Rwanda and Ghana originally. And it's just what y'all childhood brain would imagine the power of drones might be. They're extremely precise, they're extremely fast. Zipline has built up a pick and pack logistics infrastructure to keep blood cold in the drone. Just incredible stuff that you would imagine as a kid. And then that drone will fly to wherever it's needed, somewhere that might not even have roads. And it will drop the medical supplies off and they can be used within a couple of hours instead of whatever the supply chain would've been without that option. And one of the things that I love about Zipline story that they're commercializing the product now and so they're moving towards commercial drone deliveries. They're operating in a couple of US states now, but what they were able to pioneer and it took them about 10 years, relied on the freedom to basically play with airspace. And to attempt to work through all the stages of a startup, especially a hardware based startup, and get to a point through all of those iterations that they must have done, to the point where they could create something as important to the world as what they did. And you couldn't have done that in Europe. You couldn't have done it in most of the US there's the FAA here, there were rules, you can't just fly stuff around. And there's this great concept, this fascinating concept in sort of study of public policy and especially the nonprofit sector of this idea of leapfrogging. And that's where a less developed community, be it a country or whatever, due to often the lack of regulation, the lack of money being wasted on previous failed infrastructure, leaves a sort of vacuum in which creativity can actually flourish better than it might have in a more highly regulated, more highly industrial industrialized environment. And I think that's a perfect example of Zipline did of that. But the reason it's in the book is the precision of the drones. It's incredible. The accuracy they can get to is down to a couple of meters of where they're needed. They navigate themselves to and fro, they avoid collisions, they know how to land. Awesome. Definitely one of my favorites.
Juan Sequeda: Wow, this is fascinating. Because of this particular scenario, they're being able to take advantage of being in a place where I can, I have liberty or I use the airspace. So they're able to go test a lot of things out and they're gathering all that data to go see, " Hey, is this the best approach or so forth?" Because you would not be able to go do that, as you said, in different places in Europe and the US.
Maddy Want: And it was Silicon Valley science, like these people... The headquarters was always in San Fran, but where they were operating was in Africa.
Juan Sequeda: So now they're able to take that technology, what they're able to develop over there and being able apply in other places where there is regulation, but they've already kind of figured out the algorithms, the data science behind things to make it work.
Maddy Want: Yeah, exactly.
Tim Gasper: That's interesting.
Juan Sequeda: That's an interesting approach. Like, " Where can I go do something where I have a little bit more freedom to go learn and harden this out?" And then it's like, " Okay, now I can take this to another scenario where I've already de- risked a lot of the scenarios."
Tim Gasper: And it's one of those sort of win- win type scenarios where they needed to be in that kind of an environment to be able to collect the data, iterate on the data that they needed. And also it was exactly the kind of environment where this type of service could be so impactful.
Maddy Want: Right, exactly. That was an intersection there and I think that's what makes it a win- win. It's easy to imagine other variants of this kind of an initiative which might have been a win- lose. In which a less regulated or less industrialized location was sort of used as a training bed or as a test bed where maybe different standards were upheld and that being potentially exploitative. I think in the case of what Zipline did, it wasn't that. And that's why it feels like such an incredible story. It was something that was needed specifically there and that's what they did.
Tim Gasper: Yeah, it wasn't just a test bed, they were also the beneficiaries.
Maddy Want: They were the customers, yeah.
Tim Gasper: Yeah. Wow, that's cool.
Juan Sequeda: All right. You said you had two favorites.
Tim Gasper: Yeah. What's your second favorite?
Maddy Want: So the second favorite story comes from NYPD and thanks to Zach's history that we got some just incredible people to agree to talk to us just at a high level about their careers and their experience. And I don't know who else could have got that kind of access, so that was awesome. Every one of these conversations was incredible. There are some brilliant people there. Back in the 1990s, the commissioner of the NYPD was Bill Bratton and he had a deputy commissioner named Jack Maple. And they developed what is now a widely known and widely adopted data- driven policing system named CompStat. CompStat was a combination of data and process that deserves a whole case study just on itself. But the idea was very, very simple. Maple had this big map of the city and every time a certain type of crime happened, he would push a pin into the map and the pin was colored robbery, burglary, murder, whatever, homicide. And over time there would build up just a visual picture of where certain crimes were clustering over the city. And CompStat was the meeting by which all of the precinct heads were called in to commission headquarters and had to give a structured report about everything that had happened in that precinct in the last week. It was a grilling session, apparently it wasn't super fun to be presenting there and Bratton's known to be a tough guy, but that was in the 1990s and that is data that is data driven policing. And I think often today we think of data as a purely digital phenomenon. We think about data as bits, but that's not all it is. That's not all it was. That was data before there was computers. And I think this is a beautiful example of that. And in 2018 there's this sort of similar in spirit initiative that emerges where there's a data scientist in the NYPD who notices that crimes occur in patterns. And that criminals repeat crimes in multiple places because they figure out something that works and they keep doing it. It's a pain in the ass to think of a new crime type every time. And so there's this race to identify the pattern from a policing perspective, you want to identify that pattern as soon as possible so that you can direct resources to the next most likely place to get hit. You can increase focus, you can develop tactics, you can respond more quickly. Hopefully you can prevent the next crime. And for data scientist mindset that looks like, " Okay, well we've got 15 years of beautifully structured CompStat data on three levels of high level crime." And that seems like perfect fuel for a, in this case it was a random forest, where what they intended to do, what they tried to do was to create a model that could match a new crime to a potential pattern or to other crimes based on similarity. And find connections and find patterns that the human eye couldn't. Maybe some brilliant detectives out there have that just innate ability to remember like, " Wasn't there something else a year ago where this and that also happened and might they be connected?" And that's sort of what you see on the police shows, a lot of that. But in reality there's too many crimes to do that with and especially in a city like in New York City, you need a way to do that at scale. And so they went ahead and did that. They took, I think it was a decade of historical data. They wiped out zip code, gender, any sort of demographic information. It was important to do that because otherwise you would just get a biased representation of the city, which would say like, " Send all the police to all the poor places." That's not helpful. And it's not good policing either. So they neutralized anything that was identifiable and they produced three models, one to identify crimes in burglaries, one to identify crimes in robberies and one to identify crimes in homicides. And by the end they had about a 70% reconstruction rate, which means about 70% of the time the model would accurately identify a crime as part of a pattern. And they were able to train it on all of this sort of historical labeled data to say, " Is the machine going to pick up that belongs with those?" And it did that about 70% of the time. Why isn't that the way that everybody does policing today? Well, it's because of process and people and in all of the non- tech parts of data and in the book and precisely we talk about a precision system is not just a model, it is the end- to- end implementation of that model in some workflow. And it's really more a book about change management than it is about data from that perspective and the reasons why it's called Patternizer, by the way. The reasons why Patternizer ended up gathering dust was nothing to do with the efficiency of the model. It was to do with sort of everything else.
Juan Sequeda: So this is fascinating to go see how... I mean we're listening to you and okay, out of all the cool, is the technology, the data. And I love how you're just honest, no BS here is things like, " Yeah, that that's needed and that works. But what makes it not work is going to be the people in process." And I'm happy to hear that a lot of the book is really about change management. So what are the recommendations, what are the things that you're seeing? Because actually this is the topic, this is the theme of in the three years and we say what is a theme that's gone through everything, it's people and process and how they go deal with this. What are your kind of thoughts and recommendations from everything that you've been learning?
Maddy Want: It's almost like you go and identify any sort of canonical strategy book and it'll bring you through this is what the disruptor looks like, this is what the incumbent looks like, and you learn these sort of very high level concepts about markets and adoption and people. And then you just translate that into the world of data. And a good way to think about it, which is a lens that we chose to follow in the book was, " What does it mean to achieve change with data as an executive. As somebody who has authoritative power, as somebody who can set strategy and mobilize resources behind that strategy." It means something very different as a middle manager, if you're trying to achieve change through data, you don't have the same tools, your perspective is not the same and so we recommend a different approach. We've seen a different kind of approach be successful there. And then there's always the actual people who do this stuff, the individual contributors, who are the ones who create this and really bring it to life. And how do they find connection to the sorts of support and space and protection to innovate on something and then have a chance of bringing that into the hands of customers or citizens or whoever the target audience is.
Tim Gasper: When I really like this lens that you're kind of bringing in here and it sounds like is a theme that actually applies across the book, is around the disruptor versus the incumbent. And it's almost like this market lens on the whole thing. I actually want to take a little bit of a step back for a second from the stories and what's being weaved in there, and I know Maddy, you are a data leader. And do we as data leaders often have to be in the disruptor position as part of our jobs in order to be able to accomplish the value that we want to achieve out of our data initiatives?
Maddy Want: Yeah, slam dunk. No, we don't. I mean the job of data leaders is not to innovate for the sake of innovating. It's not to develop cool stuff for the sake of developing cool stuff, it's to achieve an outcome. And sometimes there is value in disrupting and innovating, and there's definitely a place for that. It's about picking your battles in what pocket of the situation around me, do we, A, need innovation, do we need disruption? And also, B, have a realistic chance of achieving that impact. There's no point throwing your best people, your smartest people at a problem, which sure, maybe quantitatively they could solve it, but in reality there's no hope that it would ever get deployed, sold, approved, whatever. And so you have to think about the environment that you're in and whether you're asking people to do something that's possible but doomed or whether there's actually a path that you can see to disruption succeeding. And then as a leader, you're obviously thinking about what are the walls that I can prevent from caving in to allow these people to do that? But honestly, for most organizations out there, nobody needs... You don't need to be in the space of innovating or disrupting. Most organizations out there could benefit from just getting up to parody with what some of the strongest established leaders in their market are already doing. And a great example of this is personalizing an experience. It's one of the most popular applications of AI in consumer facing products, is basically just the concept that you have multiple objects and you have limited real estate. Whether real estate is the size of the app or the duration of the customer's attention or whatever it is, you have to choose what to show them and why. And you have to choose the most relevant thing that they're probably going to engage with or like the most. And personalization aims to solve for that by creating an experience that feels tailored to you. The ways to create personalization in a customer facing product today are really well established. You can go Google it. There's like, " You should do some lookalike modeling. You should say people who do this also do that and that's solved for." And the majority of organizations out there could just go do that if they're not doing it. They don't need to reinvent the wheel or file for a patent or anything. Just look at what's being done around you and figure out how to do that.
Juan Sequeda: I appreciate you bringing this up. I think people are just, " Oh, I need to be innovating. I need to go do cool stuff." People are not doing data, do data thing because it's cool, it's different. I'm like, " Well, if you're not doing it, you're already kind of behind the pack and if you're not doing it well you're still behind the pack." So you just need to get to the baseline, the expectations and guess what? Let's be honest, you're catching up, that means by definition you can't be innovating. Just go catch up and get to that baseline. Later on if you're the leader of the pack, you're the one who's driving what's going to go next and people will be following you, right?
Maddy Want: Correct.
Juan Sequeda: We have to acknowledge are you a leader or are you a follower because you're not...
Maddy Want: And being a leader doesn't mean that you have to lead to the front of the market. Being a leader just means you have to achieve the impact by any means necessary. And that means maybe the strategy for the next 12 months looks pretty basic and then the next 12 to 24 months looks a little more sophisticated. You have to have in mind a journey to go on that isn't simply imagining something really cool and asking how can we make that possible ASAP? That's not how organizations achieve change. You have to build momentum, you have to build consensus, you have to prove value at a small scale, you have to earn trust. It's old school stuff in that sense, like how to win friends and influence people kind of book will tell you big change implemented suddenly isn't received well.
Tim Gasper: Yeah, yeah. And some of the biggest impact that you can have can be incremental. It all depends on the situation and it's not all just disrupt, disrupt, disrupt. For those that are listening, you don't have to figure out what your drone strategy is going to be in order to take your data projects to the next level. It could be more simple than that.
Juan Sequeda: So now I'm thinking is do you have any stories where they were not successful that we can learn from or anything you can talk about because it's like, " Oh, you were doing all these things to try to accomplish this and you didn't make it, you should have not done that."
Tim Gasper: Lessons learned.
Juan Sequeda: Or differently or whatever.
Maddy Want: Yeah. Well, I mean the NYPD one that I was just telling is a non- successful story that that project got shelved. But for the sake of some variety of the stories, there are definitely others. There was actually an NFL comparison that we did, this was long before I ever dreamed of working for Fanatics, where we'd compared the philosophy and approach to analytics of two different NFL teams within the league. And one was the Cowboys and one was the Ravens and it was a Moneyball esque story, but it was sort of a slightly modernized version of that. Where the Ravens really embraced analytics to the point that they had an analyst sitting right next to the coach advising him on which plays to call, feeding through decision support effectively. And the stories about, the team did surprisingly well in their division that year. And now that team is known for really having embraced analytics as part of sports strategy for themselves. Whereas the Cowboys took a little bit more of a cowboy approach. Their former coaches quoted as saying, " Yeah, no, we don't need any of that data. We're doing fine. We've been doing fine. We'll continue doing fine without any of that junk. Thank you." And then of course he's no longer around. They didn't do great that year. Not saying that's causal, but definitely it's an interesting correlation and it's interesting to look at two different players in the same industry effectively embracing or resisting the incorporation of data into strategy and how that went for them overall. I don't know, are you a Cowboys fan? One, I'm sorry if I offended you.
Juan Sequeda: No, no, no, no, I'm not.
Tim Gasper: No Cowboys fans here. inaudible.
Juan Sequeda: No, this brings up, I think you working at Fanatics and stuff, is there any interesting more sports data stories that you can go share?
Maddy Want: A million, most of them I'm not allowed to, but what we're doing at Fanatics is building towards being the place for sports online, the one and only, the destination. And Fanatics is a very well established, very well known brand in America and venturing into sports betting and collectibles. And live as we are this year is all about new offerings to the customer and in many cases, new offerings to the same customer. We want to be a part of people's lives when they think about sports, they're referencing something that we're offering. And the data challenge there is obvious, it's create a comprehensive picture, understand what actually motivates customers, who our customers are, who they could be, what markets are we in, what markets can we move into? And for anybody who works in data, that's a challenge of synthesizing databases and access and policy. And there's just years and years worth of plumbing type work before you can talk about that dream of the crystal clear customer 360 view. And for an established company, fanatics has been around for a long time, whereas Fanatics Betting and Gaming is brand new. We have a beautiful comparison of strategies and technologies from years ago and from right now, and it's my dream job for sure, for that reason. I think it's just a once in a lifetime chance to do a data strategy from scratch. But also, I'm coming to love the sports aspect and that's not what I went in for. I didn't grow up around sports too much. And despite being Australian, I think the Wallabies were the only thing that sort of motivated my family in the sports area. I've come to love it because I've been to a bunch of games now and I've been around fans and I understand fanaticism more than I did. And I understand a little bit more about the value and the sense of community that being fans of a team gives you, I didn't understand that before this job. So it's been fascinating from that perspective. But for me, at the end of the day, what drew me to the company and what I'm most excited about is the data challenge.
Juan Sequeda: Now, what I love hearing from you is the focus on the customer, understanding who the customer is and what you want to go, and selling, upselling and so forth. And then you really need to understand that customer. You need to put yourself in the shoes of the customer to figure out, " Hey, what does the light actually mean?" And then you can connect the dots around that. I'm always curious about this from data leaders. You just said that you're kind of in a new company even though Fanatics has been around for a long time. What is your experience about setting up a data strategy in a new place versus setting up a data strategy in an older place? How do those things differentiate or what overlaps or not?
Maddy Want: Inheriting, inheriting technology, inheriting established teams, inheriting historical data can be a real advantage because you have people that know what they're doing. You have technology that already functions and you have history and it's just shocking having no history as a data leader, just imagine you had zero history. That's a crazy place to start from, but that is a change management, again, that's probably the key word of this podcast is that's a change of management. If you're coming in to try and achieve something that's not currently being achieved, then you're going to implement changes of some kind. And whereas coming in clean slate, that is purely a strategy challenge. It's, " Where do we need to go? Who do we need to do that? What do we need to build to do that? Why should we take risk? Why should we avoid risk?" And at a high level, my philosophy is and always has been, make solid partner decisions and invest in that relationship because they'll grow with you and they'll help you. Don't invest resources in building or rebuilding things that are not IP for your business. Do you need to build a CRM engine? No, that's solved full. You should just buy that unless you're a company that sells CRM engines. I do still see a lot of companies building where they could be buying a lot and spending the time of precious engineers building stuff that isn't unique, isn't differentiating, isn't competitive. So I try to focus that way. Where there is a risk portfolio, obviously being in a regulated environment, there are very high standards to be met about how to operate the business. And if any of you know about the gambling industry in the United States, it's a state by state business. Different state regulators, different partners within each state to work with. It's not a nationwide solve. And so the infrastructure that we build and the way that we structure our data and our databases has to be very secure and segmentable. It's good, it's a great challenge. It's forcing us to invest in governance, security, et cetera, upfront that in a different industry we wouldn't have needed to be this sophisticated on this early in the game.
Tim Gasper: That's it. That's really interesting. I honestly wasn't expecting you to pose as many challenges as you had around starting with a clean slate, but I come away with what you're saying here, actually seeing quite a big challenge there versus having an established... I mean folks always talk about the challenges of like, " Oh, I went into an organization and they weren't doing things and I had to change the direction." But at least there was an established approach and people had a way of doing things and you could leverage on that. And actually that clean slate can be extra challenging because you didn't really have anything to work with and you got to choose what order am I going to tackle these things. And then also you mentioned about security and things like that, you kind of knew what use cases you wanted to address and there was no established infrastructure for that yet. And so you had to tackle that from the beginning.
Juan Sequeda: I would consider that opportunities for clean slate kind of situations are just recent or in the last 10, 15 years probably. I don't know, but what's going through my head right now is like, " Okay, we defined kind of two things in the old and the new." In the old you're like you said it's change management, you got all these existing people, teams, tools to go do. And then in a new place like, " Oh, it's a clean slate, it's focused on strategy and there's all these different... the challenges are very different." I mean obviously there's the same challenges, but the type of the ways you would deal with those challenges are very differently. So I'm just thinking about like, " Oh, I'm a data leader." " Well, okay, perfect. What type of data leader, where's your experience? Are your experience in the older places where you have had to deal with change management or you have experiences in the new places you deal with clean slate. Because that really shows where your expertise can go." And I think the clean slate data leaders are probably going to be fewer because it's more of a recent thing. I don't know, that's what I'm considering right now. I don't know. What are your thoughts?
Maddy Want: I'm spoiled because I've worked in data only in a cloud world and sometimes because of the title I get invited to data leader round tables and dinners and things like that. And I usually look at the attendee list and often it's company, people from big names like Proctor and Gamble type companies, real American giants. And I think about, " How much am I going to have in common with any of those folks?" Not much-
Juan Sequeda: inaudible the profit part.
Maddy Want: ...not much. There's being a digital native and there's being an existing company, especially behemoth that has a lot of data and needs to use its data well, and the challenges are so different and the technologies are different. I'm uncomfortable if it's not the cloud. I'll do on- prem when necessary, but really I'd rather not. And everybody can be called a VP of data, but what you're actually doing in the job depends so hugely on when the company was begun.
Tim Gasper: Yeah, that is interesting. And now that we live in this world where you could technically build your entire data infrastructure purely made out of cloud- based and modern oriented tools, it's a very different landscape. And to some degrees I imagine, and feel free to refute this, I'm curious if you agree with this or not. That maybe the job of starting from scratch, although it's hard still, definitely hard, it's maybe easier than it's ever been because of the ability to leverage cloud and modern tools?
Maddy Want: Oh yeah, no refuting for me, especially if you've been around these tools and the industry landscape, you've worked with partners that you trust. In my case, the big difference that sort of helped me sleep at night with this job, that I was able to convince one or two key people from prior roles who I've worked with and who I trust to come and do this journey with me here. And that, just the anxiety level. I thought, " Oh, okay, so- and- so is here. We're going to be good. I know exactly how we're going to approach this." And there are some things that you don't need to reexamine too closely and you can just get it done and move on. And then there are other problems that are brand new, super ambiguous, and you need to spend a lot of time on them before you can make a move that you feel confident in. And having precedent, meaning having precedent doing this kind of challenge before and having a team that you trust is the difference maker.
Juan Sequeda: So before we cut out to the next things, final question here for you is, I'm curious, where do you report to and how are you seeing different data leaders reporting to? What does that mean? Always curious about this.
Maddy Want: This. I love this question. So I report to the CTO.
Juan Sequeda: Okay. And then I assume the CTO then to the CEO.
Maddy Want: Yep. And if anything, that was the second most important thing to me about this job was I've seen data reporting into all sorts of funky places and I've seen data split up and reporting into multiple different places more commonly. And in a company where you've got data analysts reporting up to this CFO and the CMO separately. You've got data engineers sort of smushed in with cloud platform engineers and not really given their own space. Governance is sort of a weird uncomfortable side function for InfoSec. And how are you going to do anything in that environment? The cost of collaboration is just exorbitant. And if you put a singular leader in there and say, " Okay, you go in and make change in this environment." They're going to be just cuffed. What I loved was the opportunity to lead data end to end, every part of it. And say like, " Okay, I've got engineering, I've got analytics, I've got governance, I've got data science, I've got ML." It was really just like a blank check. And that has so many of the types of problems that I'm familiar with, the throwing over the fence of issues, and this group did something I wasn't expecting type issue. They just don't happen because we're all in the same chats, we're all in person together. Sometimes we meet up and it's our teammates. It's not some other foreign group that we have to go coordinate with. It's just us together. And that's been huge as well. And now I think on the theme of me being spoiled, I don't think I would work somewhere where that wasn't true now. You got to be able to make people decisions and if you are supposed to be a data leader, achieving some kind of impact and change and you can't make the right people decisions because people are all split up across various different leaders is going to be super hard to have an effect.
Tim Gasper: Just before we move onto some of our final questions and things here, how much do you perceive that a data leader like yourself can... Imagine you were walking into a different situation and things were disparate or separated or reported to the wrong place. How much do you feel data leaders can impact that or change that once they get into that role? Or do you think that really you got to decide if this is what you want to tackle before you enter the chair? Kind of like you did.
Maddy Want: I've seen it actually in a recent job in the last five years. I saw just an excruciating beautiful example of this, of somebody who was brought in to a company, given a super high title, but very few people. And said like, "We are not data driven today and we got to be. Go. And we're going to pay you a ton of money and go." And what I watched this person do over the next couple of years was just brilliant. First of all, do the must- dos and earn trust. Spend a full year doing that. Build the reports that don't exist, have the meetings that don't exist, hire the analysts that don't exist. Take the pain away, take the thorn out of someone's side, probably the person that hired you and provide them some leverage and let's all just be calm. And then over time, that person built trust and they said, " Hey, I actually think we could be doing this a little bit differently or I actually think if we invested in that, we might see this kind of return later on." And started just making very gentle sort of scope and resource asks, " Let me try this, let me try it." And then they would succeed at that. They'd earn a little bit more trust. And now that leader runs all of product and data. And because they were just so incredibly effective at fusing data into the product and the business strategy, they earned so much trust. It took years. And that is now my model for how to move in a situation like that.
Juan Sequeda: This is a beautiful statement right there. I love this. This is minute 49 that you said. Now something that I, once... I hear what you're saying, I'm like, " This is a long term." You don't do everything you just said in one or two years. Well, at least I don't think you can be able to achieve that. But then you hear the whole, " Well, CDOs 10 years, only 18 months." And stuff like that. So then it's like, "This is what you should do, but then we're not seeing this in reality in the market. There's incentives that aren't well aligned." There's a mismatch right now.
Tim Gasper: Well, I get the feeling that what you mentioned, that story there is not the normal story. And I think that there's something to be learned here for data leaders that are trying to make impactful change is... I kind of think of the story that you went through there. And I also think about, earlier in the podcast you mentioned how to win friends and influence people in passing. And I don't think this is explicitly mentioned in that book, but one of the themes of that book is people investing in people, building trust. I think about the trust bank, and if you really want to influence change and do change management, you need to put money in that trust bank. So that finally when it's time to do something disruptive, like move people from one part of the organization to the other or bigger changes than that, then you can cash in.
Maddy Want: Yeah, it's simple stuff when you think about it from this lens, but in the moment it's infinite patient conversations. It's infinite understanding and learning and listening and that can be hard to do on a day- to- day basis. And just seeing it done every step of the way was very enlightening for me.
Juan Sequeda: Well, we said we can talk for hours and I now we're like, "We're going through more and more stuff."
Tim Gasper: I wish we could go into more of the book stories too, but for all our listeners, definitely go check out the book and all sorts of stories that you can check out there.
Juan Sequeda: All right, so we're going to do this segment, the AI minute. You got one minute to rant about AI, anything you want, go.
Maddy Want: Okay, I got it. It's public literacy. As a policy person and a policy thinker, I spend a lot of time thinking about to what degree the understanding of AI. And most recently, LLMs has sort of permeated public discourse and I can see, or I fear that it's not enough, it's not sufficient. And there are countries, New Zealand is a great example, I think they did a brilliant public education campaign about, I think it was actually privacy recently. I don't think it was AI specifically, but governments can be investing in this. They can be helping people understand what does it mean to use an LLM? Is an LLM going to tell you the truth? No, not necessarily. People don't know that, they don't understand that. And these incredibly powerful technologies are being pushed so fast and they're so exciting and so awesome. And as a public sector as how are we supporting citizens to make use of these tools, is woefully far behind. And it's easy from the outside say, " You should do more, but you know, you should." So that's my rant.
Juan Sequeda: That's a unique take on that, I didn't inaudible in that case-
Maddy Want: Something actionable.
Juan Sequeda: Yeah. All right. So let's run into our lightning round. We got four questions. Some are yes or no, some are a little bit of... you got two options here. I'll kick it off. So in the stories you collected, were they just basic statistical approaches that were at the center of success, or were there advanced techniques and advanced AI at the center?
Maddy Want: Is this a yes, no question?
Juan Sequeda: No, this is the one or the other, right? So those basic statistical approaches or advanced AI techniques?
Maddy Want: More towards the latter, but heavily also... Some of these scenarios were just back of the napkin math. And by the way, I'm hijacking here for a second for a tangent, but that's also a big part of the point of this is to say, " You don't need to be doing ML. Sometimes linear regression is just fine, just do that."
Juan Sequeda: Amen to that.
Tim Gasper: Awesome, yeah. All right. Second question. So sabermetrics, Moneyball, betting and gaming, the sports world is kind of seen at the center of a lot around data these days. Have you been impressed as you've stepped into this world by what's been going on there? Or the flip side of that would be actually a little bit underwhelmed, whatever you could say.
Maddy Want: I would say the world in gaming specifically, the world of odds and probabilities is incredible. And decades of work has been done to get this industry to just a state of implementation, of scaled probability tools that I don't know where else you find that maybe in exchange markets of some kind. In terms of the adoption of data as a part of strategy in leagues, teams, sports betting operators, sports companies, et cetera. That has been, certainly not everybody has come from that background and they're beginning to incorporate it. And it has sometimes been surprising. You're a billion- dollar companies, billion dollar industries out there that just function on less data than you'd think.
Juan Sequeda: Yeah.
Tim Gasper: Interesting.
Juan Sequeda: All right, next question. Can an organization treat data like an asset if they don't have a clear data leader?
Maddy Want: They can. The two people that I think can sub in effectively are the CPO and maybe the CMO, obviously the CEO counts. Sure, fine. But in terms of people who might spot the asset value, I can imagine the CMO can think of creative ways to trade, augment, apply the body of data that they have. And then the CPO obviously is thinking about how to productize that and get it into the hands of customers. So I think you can, I think it'll be harder, that's just job security.
Tim Gasper: Just before we go into question four, I want to veer into a little honest no BS territory here. You omitted CIO and you had omitted CTO, I assume that was purposeful?
Maddy Want: CIO's an old title. I don't know if people even have that anymore.
Tim Gasper: Fortunately or unfortunately, yes.
Juan Sequeda: Actually now I'm curious to see digitally native modern cloud companies who do things when they have the CIO.
Maddy Want: That sounds very-
Tim Gasper: Digital officers and things like that now, right?
Maddy Want: That sounds very oracle to me.
Juan Sequeda: Yeah. Last lightning round question. You go.
Tim Gasper: Yes. So last question, Maddy is, are we trending as organizations towards, and I'll you use a very particular word here, more productivity with data? Or is it getting worse before it gets better?
Maddy Want: No, I think it's getting better.
Tim Gasper: Okay.
Maddy Want: I do.
Juan Sequeda: All right, Tim, take away times. Take away.
Tim Gasper: Taking away with my takeaways.
Juan Sequeda: Take away, kick us off.
Tim Gasper: Oh my gosh. So much good stuff today. So we started off with this phrase, which is, " The criminal under leveraging of data." And any organization you said that's not treating data like an asset is criminally under leveraging it and stop committing crimes everyone. Leverage your data, use it as an asset, treat it like an asset. And as a digital world now you mentioned hopefully we're moving past the confusion of what data can do for your business. There are so many creative possibilities. It's not all only about business reporting and dashboards, and you answered that. What does it mean to treat data like an asset? You mentioned that it's closest to an accounting perspective. Which side of the balance sheet is it on? Earlier in your career you mentioned that data was more like a cost center undervalued, but now as we go forward, it can lead strategy instead of being a liability it can be an asset. Data driven data literacy, those aren't end goals, those are must do. Those are table stakes. The goal is to do something of value for people and for customers. Delight your customers. And you mentioned about your book. Your book was four years in the making. Your co- author Zach, was formally affiliated with the NYPD and leveraged his network and experiences there. And you interviewed folks for two years and then spent another year editing, editing, editing, and pulling together this really awesome collection of stories of how folks have leveraged data. Sort of the leveraging of data, and perhaps in some cases some criminal under leveraging or mistakes that have been made along the way. So I think that's awesome for folks to check out as a follow- up to this. Your favorite story, two of them, one was Zipline, which was pioneering drone delivery of blood and medical devices and services to remote parts of Africa. And how even though they were a company headquartered in San Francisco, they were able to operate in an environment in Africa where they could experiment, they could iterate, but they didn't do it in a way that was a win lose. Where maybe they were taking advantage of the situation there, taking advantage of people. It was actually in the service of those people, those people were the customers. And so it was a win- win. And so that's a great story of leveraging data to make these highly accurate drones that could not crash into each other and things like that, while also helping people. And then you also mentioned around the NYPD, how back in the nineties I think it was, they created CompStat where they were taking the data around crimes happening in the city and being able to keep track of history and trends. And then in 2018, there was a similar in spirit initiative Patternizer, I think you mentioned. Where they looked at the patterns across that history and when a crime happened, they could kind of predict if that crime was going to repeat to very high accuracy. And unfortunately, that project was ultimately shelved, which I think might be a little bit of a handoff to Juan. Juan is going to talk about here around change management. So Juan, over to you.
Juan Sequeda: Yeah, and like you said, your book, it's actually a lot about change management. I think that's a really important part. You got to take the market lens, " This is what a disruptor looks like, what the incumbent looks like." Understand the three levels, executive, middle management, individual contributors, like, " Where do things go?" And I love your very honest no BS take here is like, " Hey, data leaders. You're not there to innovate for the sake of innovating. It's all about outcomes. Pick your battles and understand where do you need to innovate? And sometimes guess what? You probably don't need to innovate. You really need to be thinking about that environment that you have. Is it possible to go do this, but will it be doomed because you won't be able to deploy it? So don't even consider doing that." And sometimes many organizations are just so behind the pack that, " Oh, you can benefit is just getting to parody." That means that there is no innovating disruption needed right now. You just need to play catch up, go focus on that. Because being a leader means that you should achieve the outcome. And this may mean that the next 12 months is just doing the simple catch- up stuff. Maybe later you get a little bit more innovative, but you got to do that basic stuff. We talked about, what is it between the older companies and newer companies? So in defining strategy, if you're in an older company, you're inheriting people, you're inheriting teams, you're inheriting tools. It's all about change management. While in a new place you have a clean slate. It's a strategy challenge, but there's a lot to consider there. Where do we need to go? Where should we take risk? Where should we not take risk? Who are the partners that we're going to go work with and that we're going to make solid decisions? Where should we be investing or not investing, doing things? Don't invest in things that are not part of your IP. Invest in security and governance upfront. And in your case, you said it kind of spoiled because you worked in that cloud world. And as data leaders, you understand which world do you live in, and that's how you'll understand how you relate to other data leaders. And then we finalize with reporting, where does reporting... and you report to the CTO gets up to the CEO and, but we see that reporting can happen so many different places. Especially if you see data reporting to the CFO and CMOs and governance, reporting to InfoSec stuff that makes it really, really hard. And there's always a preference to lead the data from end to end. It's kind of that blank check. But if you don't have that possibility, you just got to start with the must dos and generate trust. I like how you said, " Just do the things that don't exist that need to exist. Build the reports that don't exist." Take that thorn away from the person who hired you, start making trust and then you can ask for some little change. What do you think about that? Generate more trust and that's how that cycle continues. How did we do?
Maddy Want: Beautiful.
Juan Sequeda: Anything else?
Maddy Want: Wow. Recall 100.
Tim Gasper: Who need needs ChatGPT?
Maddy Want: We got Juan and Tim.
Juan Sequeda: All right. Well to wrap this up, let's back to you, three questions. What's your advice about data, about life? Who should we invite next and what resources do you follow?
Maddy Want: Okay. My advice is something that it took me way, way, way too long to learn. It is you do your best work when you are getting your best rest.
Juan Sequeda: Amen to that. When I was in grad school, I was also sleeping not that much. And then I had a shift. I started sleeping my eight hours and I realized, " Oh, I'm kind of being more productive."
Maddy Want: inaudible now. I feel smart now.
Tim Gasper: Don't just burn the candle at both ends. What about, who should we invite next?
Maddy Want: Okay, I got two people on the list. I don't know, they're both very important people. I think you should shoot your shot. One of them is Tom Davenport, who is just one of the best contemporary sort of thinkers on AI in the context of corporate transformation. I think we were lucky enough to convince him to write the forward to the book, which was just stunning. But he has also published a book recently. It's fascinating and it's all about, again, just building that bridge between the people who are pioneering AI and sort of the rest of the world. And he's very, very good at that. And the second person I would say is somebody who I started following a couple of years ago, whose perspective I really appreciate. She's called Timnit Gebru, and she was well known for her prior role at Google. But she's been a researcher and I think of for a long time. And just the perspective is, it's her perspective is generally a great reminder of things that can often be overlooked in the world of the excitement of innovation. And we have LLMs that can do incredible things now. For example, just yesterday or today, she had posted something about, the workers who do content moderation training for the models that power many of the products we use day to day, Facebook, et cetera. They do content moderation, like content labeling, yes, no, appropriate answer, inappropriate answer. And they filter out some really disturbing content and prevent it from arriving on any of our screens. And there's a company who provides these services in Africa as a few companies actually. And those workers unionized because they weren't getting adequate sort of rest, treatment, compensation, rights, et cetera. And it's just this reminder that these lovely tools that make our lives easier, it takes a lot to get them up and running and some of what it takes is really tough stuff. And you got to think about the whole evolution of that beautiful, handy product that you're now enjoying and whether it was ethically made at every stage of the process. So she's great.
Juan Sequeda: Yeah, no, she's fantastic. We've following all her stuff for so many years and a lot of amazing stories she can tell. Finally, what resources do you follow, do you recommend? People, blogs, podcasts, books? I mean, obviously your book.
Maddy Want: Yeah, I got one great resource for you. I think this is an illegal answer, but I don't have favorites. I just snack, I just peruse. I'm on people's personal medium blogs. I'm following all of the big news channels. There's no go- to for me. I do feel like that's an illegal answer, but...
Juan Sequeda: No, no.
Tim Gasper: We've had folks say LinkedIn, yeah.
Juan Sequeda: LinkedIn, I just follow LinkedIn.
Maddy Want: LinkedIn's great.
Juan Sequeda: Well, Maddy, thank you so much. Just a quick reminder next week our guest is going to be Ben Clinch. He's a enterprise architect at BT, a British Telecom called BT. We're going to be actually live from Gartner in London. We're going to have a slight change. We usually do this live 4: 00 PM, we're going to change it next week. We're going to be live at 11: 00 AM US Central Time, which will be 5: 00 PM UK time. That will be next week. And with that, Maddy, thank you. Thank you so much. And again, check out your book, Precisely Working with Precision Systems in the World of Data. And as always, thanks data.world, lets us do this every Wednesday.
Maddy Want: Thank you for having me. This was fun.
Juan Sequeda: Cheers, Maddy.
Tim Gasper: Cheers.
Maddy Want: Cheers.
Speaker 1: This is Catalog & Cocktails. A special thanks to data.world for supporting the show, Karli Burghoff for producing. Jon Loyens and Brian Jacob for the show music. And thank you to the entire Catalog & Cocktails fan base. Don't forget to subscribe, rate and review wherever you listen to your podcasts.