[MUSIC] >> Hi, I'm Eric Dots. >> I'm John Wessel. >> Welcome to the DataStack Show. >> The DataStack Show is a podcast where we talk about the technical, business, and human challenges involved in data work. >> Join our casual conversations with innovators and data professionals to learn about new data technologies and how data teams are run at top companies. [MUSIC] >> Welcome back to the show. We're here with Edward Schenard, who has had an extremely illustrious career across both large enterprise and startups in data. Edward, I have so many questions. Can't wait to dig in with you. Thanks for joining us. >> Thank you for having me. >> All right. Well, can you give us a quick background, just a brief overview of your career and what you're doing today in data? >> Yeah, I've spent most of my career in analytics are as a market analyst. The last 13 years I have been building and running data teams and includes engineering, data science, analytics at Fortune 500 companies and startups. Currently right now, looking for the next big thing to do. Talking to some companies right now, I think it might be getting close on so. >> Awesome. So before the show, Edward, we talked a little bit about P&L ownership for data team. So I'm really excited to dig on that topic and then what kind of topics do you want to talk about? >> Yeah, the building of teams is really important to change and how like AI is changing the dynamics of the space. As well as areas of like personalization, it's interesting. It's one of those things that every few years comes back to the surface as an important topic. >> Yeah, awesome. Well, let's dig in. >> All right, let's do it. >> Edward, I'd love to start out by talking about your experience as a data leader in the large enterprise. One of the things that you've talked a lot about publicly and I'm sure privately as well, is this concept of P&L ownership by the data leader. So the first question I have is, you get a ton of pushback on that. Well, actually, before we start, explain that concept to us. Why are you so big on P&L ownership for the data leader? >> Because to me, I think the data leader is really a business leader in a lot of ways. And just like any business leader, they should own a P&L. They should be looking at profit and losses. They should own the tools and the relationships around those tools. That then makes them a true business leader because what happens is when I was an analyst, you're spending your days creating the analytics reports and finding the insights needed. But then as you move up into a leadership role, that becomes a very different role. And I think that's something that a lot of people don't understand is the leadership role is not really a role that represents the analysts, the data scientists, the engineers in their data data function. They're the ones helping to expand the market, expand what the team does. So P&L is really important because you need to show like, hey, this is how our team is impacting the organization. This is how we are engaging customers. This is where our money is going and why we're spending it that way. And if the leader isn't able to articulate that, then the team really does not look like a serious team to the rest of the organization. >> Yeah, so from a cost standpoint, nobody's going to fight you as a data leader to own the cost part of the tooling and infrastructure. But from the profit standpoint, I'm curious to dig in on that. What are the things on that side? Because that's harder, right? >> Yeah, that's where the fights always are. And this is why I often take a product approach to how I run the teams in terms of we need to build our own stuff. And I've done that at all the companies I've worked at because we need to show it's like, hey, we're actually contributing here. So like in my last company, we actually were like, it was interesting. They were doing all these custom reports for customers for free. And like we had one customer, they said, we want all our reports in cases of wine, because they were a wine distributor. I was like, you know, there's time that's going to take? Why the hell are we charging them for that? >> Right. >> Yeah. >> We could bring in like tens of thousands just on these custom reports every quarter. Why aren't we doing that? Now all of a sudden it's like, hey, I just paid for my analysts based off these custom reports being paid. >> Right. >> And then it was like looking into other things like, well, now all of a sudden they want consulting too, because they get these reports and people in logistics often, they'll put somebody that doesn't have the experience and they're learning logistics. So then you give them a report and they're like, I have no idea what I'm looking at. Well, what's a consulting business and charge them like, hey, for this project, we'll charge you $15,000 for it. Now, some I've got another revenue stream coming in. Or it's like, we'll take, you know, there are times where we've had to take products from the product team because they were just stretched then, so we take the ownership of it. Now we're building that product down and expanding the revenue for it, working with sales. And this is where also the dynamics of the team changes because I'm just being honest here, a lot of people that work in data and analytics are introverts. And they like being in the background because they don't want to deal with people. But I'm like, sorry, you need to deal with people because you need to understand what's not in our database, but what's being communicated to sales and management. So I'd have OKRs saying you have to have X number of customer engagements each quarter. And that did really help because all of a sudden they were realizing, because I'm dealing with a lot of people that are like in their 20s or 30s, that don't have a lot of that customer interaction, especially the last four years where we were going remote. And I'm like, right? I'm like, hey, I actually feel sorry for these kids who are out of school. Like when I was their age, I'm in the office. I could watch other people and see them when they do things for them. What are they doing at Zoom call? Then when that's over, they got nothing. Yeah. Yeah. What pushback do you get on this concept? Because you've said that there's a ton of fights that develop around this. And, you know, we've seen some of that online. What pushback do you get? I get told that you don't need to own it to be the leader. I'm like, well, there's a difference between a big C and a small C for talking like cheap data officer. Small C does not own the P&L big C does because now they're being told like, Hey, what are you doing to drive revenue of this quarter? If you're not being asked that question, you're you're a small C as a chief data officer. And a lot of the people that don't want the stress of that ownership are saying, why don't really need to own it? I'm just going to build dashboards or I'm just going to make sure that my uptime on my data pipelines are good. And I'm saying, well, look, you want to have the respect of being a big chief data officer, then you need to have the responsibility to. But a lot of people want that responsibility. And that's why it's really common. How do you carve out that role in between, like, say a company has a CFO and a CIO CTO, how do you carve out that role between those two positions, like as a chief analytics or chief data officer? Have you seen it done? The CFO has never been a problem for me. They tend to be like, yeah, what you're doing is different. Yeah, yeah, the CTO CIOs. Yeah, they're tends to be friction there and with their teams. But what I like to point out is like, you know, if they're running software engineering teams, that's a very different mindset. Software engineering is deterministic. It's very much like, hey, we have our process here. We go through it. Data and analytics is very much trying to understand what is the problem and then the solution and then, okay, then you start to have some similarities with software engineering after that point. But it's a lot of work that takes place before that. And that problem solving is where I carved the space out because the mindset of your data engineer, data scientists and analysts is different from the software engineer. Yeah. And what they do and what they focus on is very different and even the tools being used. And I think that a lot of places, they place the data team under the CIO CTO because they think, well, they're both working with data. So it's the same, right. But no, the mindsets are very different. Often the problems are working on are very different. And just the way the teams communicate and get work done is very different. That whole issue is not solved. And I don't think it's going to get solved anytime soon until more companies realize really those teams are not the same. It's like a sales and marketing is the same because the cost of customers. Well, you know, if you go to any big company, you know, that's not the case. Yeah, over the past decade, have you seen any shift in the mindset around that? I mean, data has always been important. But I think, especially, you know, just with some of these, you know, developments over the last decade, both in terms of infrastructure, right? So it's cheaper to store a lot of data, which means you have more data to work with. You know, there's, you know, the tooling has changed, but also the mindset around it. Right. I think that there's also been, you know, and of course, the crochet is, you know, Jen AI has, you know, highlighted the importance of data, right? When of course, it's really been important all along, but in terms of understanding the unique nature of working with data and how to actually turn it into some sort of value for the business, have mindsets across the organization change it all? Or do you think we're still in a similar place to 10 years ago? It's changed, but they're actually not in a good way. Oh, interesting. OK, I got to hear about this. So if you talk to, you know, like I go to there, there's like this data leaders group here where I live in the Twin Cities and you'll meet up like once every three months. And you can see like this big difference. And there's also another bigger one called mini analytics, which I used to be a part of they put on data conferences. But once I got married, got kids, I couldn't dedicate that kind of time. But I sometimes go to their networking events and you'll see that the people that have kind of been in the trenches for a number of years, I'd say, like once you get that like seven year mark, you start to change your mindset. And you're like, yeah, there's a lot more nuance here than before. And yeah, I was like that too when I was younger, you know, I used to think like, hey, if I follow the process correctly and I do the analysis right, then clearly you should listen to me, right? You know, when I brought that up to my dad when I was starting out, he did that same laugh too. Took me years to realize why why he laughed at me and what he said was, well, there are times yet you're right, but there are other times I would rather trust somebody who has 20 years experience and knows the nuances. And that's just it. It's like, if you've been in the trenches and you start to see the nuances, you start to realize, hey, some of the stuff that we think is true is completely wrong. That's what I see when I go to these meetings. It's like the ones that have been in there a long time in the trenches, they're like, yeah, how I thought in 2015 is totally wrong today. Right. Get into the news. That's why I'm saying like, but when I talk to the younger kids and like my last job, I had some of them critique me saying like, hey, you know, Edward's looking at the strategy and he's looking at like talking to customers. And I just want to know like, Hey, how do I like employ this model here? And I'm like, well, is that even the right model? You went on Google, looked it up and decided, well, I'm going to pick that one on the list because I like it. And I'm sitting there saying, well, does this fit our customers needs? Does this fit the company's needs? How who's going to maintain this and who's currently maintaining this? You haven't answered any of these questions yet. Yeah, it's the thing is like the deep thought, just it's kind of lacking. And I don't blame anybody for that. It's just kind of like our use of technology. And like, you know, you go out there and there's TikTok videos, YouTube shorts, this shallow thinking and this ability to think deeply is just diminishing over time. And you can see it in the books up there too. I rarely buy a new book on data and analytics because they're just so shallow. I'm rereading my old books from like a decade plus. I'm like, why am I getting so much more information, better thought in this book? That's from 15 years ago that a book that came out six months ago that in theory should have more details in it because we've had more experience over that time period. Right. Yeah. I think there's another vector you mentioned wrong thinking, right? I think I think there's this other vector of like, I'll call it like usefulness. Like, hey, what can we safely ignore and how can we focus on the right things? Because I think people can really get caught up on like, oh, this is right. This is wrong. This is right. And the question might actually be like, is this even relevant or useful? I've seen that a lot. Yeah. Yeah. Yeah. Yeah, that goes back to like, well, who are you really doing this for? And unfortunately, I've seen like, particularly data scientists doing this where they're really patting their resume. You know, I've gone into companies where it's like, I remember when deep learning was a thing. And all of a sudden, like, you know, I said, see a tropics and also a bunch of data science was like, we got to do deep learning. Like, why? You know, the problems we're solving here do not require that. And then you just like, well, you then you like started talking to them. They're like, yeah, well, you know, I heard some guy who knows somebody who knows this person making seven figures doing deep learning. Like, yeah, I'm attached to that. It's just yeah. I'll never forget really early on in the show. I mean, this is like three years ago, probably that was probably one of our first handful of guests. Brooks can probably tell us he can look it up. But it was the CTO of this company called Bookshop, which is sort of like a, it's an online book retailer. And they do a number of different things that are kind of cool. Anyways, different from Amazon, different from Amazon. Yes. No, they like give money to independent bookstores for. Oh, it's cool. It's really neat. And actually it's like pretty gigantic now, I think. Anyways, we're talking with this guy about their stack. And he's like, I'm going to be really honest. Like it's really boring. Like it's very boring. It's pretty simple. Like that's it. You know, like we have a couple of pieces here and like, you know, like there's one really hard problem. Episode eight, September 30th, 2020. There we go. Brooks. Thank you very much. Um, Mason Stewart's actually a guest. Great episode. If you want to go back into the archives, but he's like, you know, we have one really challenging long tail data problem to solve around some sort of classification ID or something, which is data that they got from, you know, book classification, some public, you know, classification system. That's just a nightmare to deal with, right? Because of all the, you know, it's a public data set and whatever. All whatever those things are. And he's like, that's kind of hard, but like we figured it out and he's like, it's boring, but it does the job extremely well, you know, and nuts of the business needs. So anyways, that really resonates. And it just always reminds me of that like, you know, I'm not going to, you know, I'm not going to do something fancy because we don't need it. Like we just don't need any fancy things. Yeah. You know, when we built the personalization platform and Best Buy, you know, we were like updating our catalog like every 20 minutes. And for most products, that's a total waste of time. Mm, no, when you get the new Apple, whatever that comes out. Yeah, that's what's useful. We're doing, you know, like holiday season was funny. When I went to Target, their personalization team was, you know, struggling on some stuff and they were like, we're going to go to solid state drive servers and update every six seconds. I'm like, what the hell for up to the minute, you know, instant results. I was like, well, okay, but your customer works in human time. So they're not going to make a decision in six seconds on do they buy that beach towel or not? And the cost that you're going to put into that is just not worth the squeeze. So it was just like, you know, but they wanted that. They were so into the tech. They were in love with the tech, not the customer experience. Yeah. Well, and I think you just made a great case for the P&L ownership right there. Because because if you have P&L ownership, then that matters to you directly, right? Like the cost benefit, but if it's, if you're just part of a group, part of a cost center, then it's like, yeah, I don't know. This is how much it costs. Yeah, like, you know, it was funny. So I'm very much into telling my teams, like everything I know when I say, when I was a best buying, we're going over the roadmap. And of course, I tell the team first, and one of the developers, you just raise his hand and he goes, what's ROY and why do you keep mentioning it? Hey, I have an MBA. So yeah, okay. I'm very, yeah. I'm thinking like, well, everybody knows what ROY is and why it's important, right? Because, you know, it's drilled into me at school. But here I am dealing with somebody who maybe he took a bit, you know, general business class in school, and that said, and so explain it to him. And he's like, thanks. Now I actually know why we're doing what we're doing here. Oh, yeah. That's huge. Awesome. Can we dig into the Best Buy project a little bit? So that was, you know, before the show, you're talking about how there was very little in the way of, you know, sort of true personalization. And then it became an extremely large source of revenue. But just take us to the beginning. When you say personalization, I mean, that's such a hot topic and, you know, marketing, you mentioned that it comes up every couple of years of like, okay, personalization is like the new thing, right? And it's like, well, you know, like it's been around, you know, since before computers. Yeah, since the beginning of time, you know, since instead of cafe, somebody wrote your name on a cup. Yeah, yeah, it was, I'd always sell a team. I was like, we're actually bringing what's all back. Mm. Like we're trying to recreate when the shop owner knew their customers knew, like Mrs. Smith comes in on Tuesday and all the way through self a chocolate. Yeah. Yep. That's what we were recreating. So personalization to me, I actually had somewhere I've got this presentation, like 18 ways to personalize. But, you know, for most companies, it's recommendation engines, which isn't true personalization, you know, the email marketing stuff. And then what we were really driving towards was we actually had three levels of personalization, which was crowd driven, persona driven, and true one-to-one personalization. When I started at Best Buy, it's 2011. Okay. They were using rich relevance and basically what happens with a lot of the algorithms is they got what we call flatlining. You'll see this period of revenue growth and then all of a sudden it just flat, flat. So, okay, though Best Buy had hit that and they were like, Hey, rich relevance. Help us get more and rich relevance like, Hey, our stuff's all proprietary. So they were like, Hey, okay, we'll just do our own stuff then. Right. Yep. Oh, that's how I got hired on to run that. Now when I started, it was literally just a team of me. Oh, we worked. I worked in a group called Emerging Technologies, separate from IT that matters down the road when I got there, IT is like, what are you building? I was like, well, I'm going to go out, research what everyone's doing and come up with my own, you know, approach to how we should do this. It's best for Best Buy. So I was able to like reverse engineers, other companies at the time, the guy who ran Amazon's personalization platform had a really big ego. I found a forum where he like to hang out and I intentionally started saying stuff wrong. So he would correct me. He literally like told me how personalization is run at Amazon. That's amazing. That is really funny. And so what I did was I was looking around and I realized, Hey, we need like some kind of big data thing played with like, you know, MongoDB, you know, React, RevitMQ, and then settled on Hadoop. Cause that was the only thing that did not break when we threw our test data set, which was throw the holiday testing data set at it and see if it jumped. Yeah. Yeah. That was the only thing that did. So we built a really simple stack. Now at this point, now I got on in a business analyst and a database engineer to help me build this. And we went over to the data center in, you know, across the street. And just on commodity hardware, built it out ourselves. Now there wasn't like, you know, any Coursera courses or stuff that nobody was out here, you know, we couldn't even get like, you know, Hortonworks or Cloud Dara reps to come out to visit. It was like nothing. I basically bought a book on Amazon by this professor at Stanford on like building big data sets. And that's what I used to build it. Nice. Wow. We got it working. And I was like, Hey, I'm going to do all open source. That was sacrilege at best by the time you don't do open source. But I said, Hey, I've got stuff that works on commodity hardware. The data centers about to get rid of all these servers. Why you just give me the servers? So you don't have to pull them out. Data center guys were like, yeah, we're all good with that. And then IT was like, no, you can't do that. You can't use open source. So they brought in SAP Terra data to bid on it. It was actually kind of a good thing for me. Because these guys read 20, 30 million, 18, 20 months, just to build the big data structure. And I'm, I just came back and said, give me one quarter. We'll be ready for production. Wow. And I said, give me half of what they're asking for. So they did. And yeah, we were ready in 60 days, had our first algorithms out there, you know, by the end of that quarter. And as we started and gave me the most restricted view of how I can credit a sale to our products in the session. You had to have bought the product. So if you came back like two weeks later and bought it. Oh, wow. That's really rigid. Yeah. And hired some data scientists. And again, well, now we're into 2012. And I'm like, I have no idea what data science is. You have to help me figure that out. And we were doing that now. It's the year of the three CEOs at Best Buy. So in 2012, they had three seat notes. And that's important because they basically were ignoring us. I was like, Hey, everyone's worried about what the next CEO is going to be interested in. I'm just going to build this. And yeah, I hired my own team because IT wouldn't help me. So I basically hired a bunch of contractors to come in to to build it for me. And we were just cranking stuff out every two weeks, which was unheard of at Best Buy at the time. They were, you know, a sprint for most teams with four to six weeks. We're doing it too. Right. Well, we could process down where we could build, test, launch, an algorithm, a machine learning algorithm in one month. Wow. Well, then we're going to start, put them out there. And I go to the call center and I see like, Hey, they could use these algorithms too. I go to Best Buy for business, geek squad, even the distribution part of the business. We're just spreading our algorithms everywhere. Yeah, and we're collecting all this data. We actually got to the point 2013. We collected more data than the rest of Best Buy combined. Really? Yeah. And what, and we're digging a little bit too. Like, when you say you deploy these algorithms and you talked about those three different types of personalization, what was the algorithm doing? And I think it was, you know, crowd, we had crowd persona and one to one. Right. So to get rid of the cold start problem, we were using the crowd driven one. So when you go online, you know, people who bought this also bought that type of algorithm, what that would do is give us a pattern. So back then we could start tied to a device and we're saying, Hey, I don't even know who this person is, but I've created persona profiles based off of different behavioral patterns I've seen and I'll match them in and start recommending products based around the persona. So we use the corporate persona that because, you know, I talk to the teams that don't that in the UX team and cut in CX teams. And I was like, they did a great job. They'd go out, follow people, sit down with them. I mean, it's kind of kind of weird. They're sitting there like having like dinner with these people in their home. So not the kind of thing for me, but they did the real work. Yeah, they did the real work. Yeah. So I was like, Hey, I'm going to use these as our personas, because they did the good, they did great work. And you know, then they would bring us in because they had what we called we nicknamed it the interrogation room. Best Buy has one of these rooms like, you know, you seem like the cop shows, the one way mirror thing. Somebody asking questions. It was nicknamed the interrogation room. You could sit there and like watch like how people would interact with the algorithms and see how what their responses are. And it was great. So we would figure out like the customer journey once and see like, Hey, where are the points where they're dropping out and what algorithms might help them to stay in and get to that purchasing point and make the purchasing decision because when it comes to electronics, your average person looks at like 10 different websites before they make a purchase. Sure. Our own research was like, you could put like Bob's electronics on the best buy. Nobody cared. What they were interested in was a product they were buying. Now, we were also like, Hey, how do we make? How do we take it from a commodity purchase to an experience? So that's why the platform was called the experienced platform because we were trying to make it so that the act of purchasing was an experience in and of itself and a positive so that when we do the follow up remarketing, trying to get you to buy the accessories, you're going to be like, Hey, it was a good experience buying the main product. I'm going to go buy the accessories there too. And then on one to one, we were trying to get to that, like, Hey, who are you buying for? You know, if it's not for yourself, what events are happening? Like if you've got students in your life, Hey, August, we're hitting you up with the back school stuff, all that sort of thing. Yeah, yeah, the Christmas thing, anniversaries, making it a true. We're looking at your past purchases. We're looking at your current searches and we're trying to figure out your life events to see like, Hey, what's going to be getting you? Because most people, they, except for appliances, their purchasings tend to be pretty much the same across all product lines. Appliances are different because, Hey, if your refrigerator goes, you're just going in and say, Hey, I need a refrigerator. Interesting. If you're remodeling, you might be looking at it for months before you make a decision. Interesting. And so, but if you think about something like audio equipment or, you know, a TV or something like that, those behaviors, you're purchasing behavior and research behavior tend to be the same. Yeah. And then you have to look at like little nuance things too. So this was a fun one. The analytics team was struggling because New Hampshire, which has the lowest population when you look at Maine, New Hampshire, Massachusetts, during Black Friday and Cyber Monday, they would have the most sales in the store. But only on the borders. So you look at ports next to Maine and a Nashua next to Massachusetts. Those were the highest performing stores. And you know, they're scratching their heads. I'm sitting there going, well, you know, I grew up a few years in New Hampshire as well, they have no sales tax. Yeah, that's what I was going to guess is sales tax. I'm like, everybody's crossing the border because they just save themselves 10% just by crossing the border. But we were sending them emails, hey, go back to the Nashua Portsmouth store for like, if you bought a DSLR for the training class, it's like, take their own address and then map it out to the nearest. Yeah. Now it's like, hey, well, it's closer. So it's more convenient. Or like when we were looking at locations. So somebody who's in South North Dakota, they'll drive two hours to go to a store because they kind of have to. Yep. Right. Atlanta, Georgia, they won't drive five miles because of the traffic. Yeah. So, so like for the call centers, if you're talking to somebody in North Dakota, you say, well, hey, go to Bismarck, go to Fargo, if you'd like to pick that up today for somebody who's in Atlanta, but it's more than five miles. Say, hey, do you want us to ship that to your house? Yeah. So that's really hard. characterization would would come in. You're using that, that that location in like the environment in which they live in, like what your recommendation is. Was it hard to? So there's the algorithm piece and there's getting, there's, it's getting those things right, you know, where it's like, okay, you're solving this problem around, you know, you know, those cases that you just talked about to create some sort of great experience for someone. But then you have to get that, you have to sort of put that data to work in that. Okay. There's probably a website component to it because it's in session. There's probably some sort of messaging component around email, but it sounds like Best Buy was a really interesting environment at the time. Was it hard to go to those teams and say like, hey, you need to actually overhaul your email campaigns to use the outputs from this algorithm or to work with the like user, you know, website team or user experience team? Yeah. So that's where like all the corporate politics comes into play. Cause you know, again, like it's a year of the three CEOs, everyone's like buying and jock fiend for position. I'm really an upstart. So there are established teams like the dot com team that runs the main website. I mean, this is a team that takes up like two or three floors. And in the share. Yeah. Yeah. And here I've got like a small section on one floor that like, you know, if you blink when you walk past us, you totally miss us. So yeah, it's, but it's learning like, Hey, what motivates them? Mm. So like when emerging technologies, we, you know, me and a couple of others, we actually migrated on became the omni channel team because we went from just purely digital to digital and physical. And they're, you know, like with the stores. Now, if you talk to the stores, they've got two, three year roadmap. So they're like, Oh, great idea. Yeah. We'll implement that. You know, what I would do is like, Hey, I need to test something out here. I go to the store managers like, Hey, I've got something that can help you make more money than the store manager in town. They're like, yeah, I'm listening. Yeah, that's great. So what would be like a implementation for in a store of, because we all think, I think of like the dot com and the website obvious implementations. But what would be a store? Well, like the vending machines, you probably saw on like airports. That would. Oh, yeah. Okay. Well, the one thing that was what to stock in the vending machine, right? Yeah, we're doing stuff. But, you know, the actual stores themselves, the thing that was really interesting was the tablet experiment. So a lot of people, you know, again, talking to this, you know, the customer insights team and so on, they give us feedback. Like people don't trust the high school college kid on this $10,000 electronics they're about to buy. It's like, well, hey, give them a tablet that gives them like the reviews, the specs, you know, everything you'd want to look up. So then the employee can be like, Hey, don't just listen to me. Here's all the information. Oh, I just loved it. The looked at the CFO hated it. Oh, lately. So she just thought people were going to like steal the tablets. It's like, Oh, okay. Oh, they're formatted for us. And, you know, it's not like you just walked out with them. Just started using it because, you know, it was stripped down to just the best by application. Yeah, and so that was one of those things where I was like, great idea. But yeah, it died because somebody higher up had an opinion. And she was just too stubborn to change her mind. Wow. Sure. Like the fact that the customers and the employees loved it. The other one was we had this big, like, so when Best Buy had the small format stores that you would see in malls, we put this kind of kiosk machine in there. It was basically like a big touchscreen TV. I'd go to the store managers. They'd be like, don't get rid of this. We like this. This helps us a lot. Again, they were like, finance was like, nah, too expensive. Like that's one of the reasons why the small format stores died is because they were putting costs ahead of the experience. And my logic was if you create the right experience, you'll create enough revenue to cover the costs as long as you're managing the cost. But don't sacrifice the experience just for cost savings. So so in that case, what was the experience? Then you said there was like a touchscreen and like customers would interact. So because it's a small format store, it didn't have a lot of products. So what they would do, the employees would go to the touchscreen, work with the customer, figure out exactly the place in order and then they could have a shot or yeah. Okay. Yeah. Super interesting. Yeah. Well, you know, minor things like we also did the in-store pickup and curbside delivery. Sears and Kroger's were like the only ones doing that before that. But they, and things like like, if you went to like a Kroger, it was like two hours before your food would be already. Right. Right. Wait a minute. What's this experience like for most people? And it's like carry out pizza. Yeah. Yeah. Right. Let's order some pizza and go pick it up. How long does it take to be right? About 20 minutes. So why can't it be ready in 20 minutes for us to? Yeah. Out target when I went to target, they were just like doing like a brain dump out of me to like how to do that there too. I'm like, they hadn't gotten that in place before lockdowns. That would have been a very different experience for target. Oh, target's really good. I think in my opinion, they're one of the best at that experience. Yeah. Yeah. And they basically just learned that by, you know, that's how I you know, target, they basically like threw a number out. I couldn't say no to. There you go. And to close out with the best by story is amazing. But of course we have to talk about AI before the show's over. We would have failed everyone. But how do what's the conclusion of the best by story? So, you know, obviously, some things worked. Some things, you know, died, but what was the impact on the business? Oh, the, I mean, that the whole data science data. Engineering, personalization, one, when Jolie came in, who was the guy credited with the turnaround, you know, I get a, I get asked to like present what we're working on. And they're just like, you know, the executive team is like, is this a lot? I would say, yeah, everything I'm showing is live. You're like, you're the first person to show us something that's live. Wow. So the whole idea of two weeks sprints took whole open source. Lots of other teams started doing that. The idea of focusing on the customer was my takeaway, not most people's takeaway, but to me, I was like, technology for the sake of technology has a waste of time. And I saw many fail because they did that. But if you focus on the customer and their customer experience, that's what really drives it. And then the technology becomes the easy part. And that because people are not easy. I mean, most people make an emotional decision and then look for a rational excuse for it. Yeah. Yeah. But the end result was that whole group was bringing in over a billion dollars a year when I left. Wow. Yeah. I mean, from going from almost nothing and one person. Incredible. And wonder. You know, it's not your target one of you. Well, it's not just the technology. It was also the way you manage things. Like I said, I use a very emergent strategy approach. I'm very proud that a number of people who worked for me have gone out and become VPs, C-suite, managing directors, because they were in an environment where I allowed them to think critically, solve problems and get polished in terms of how they present themselves. Yep. Love it. All right. We have to fit A.I. and be free to be a show about data. All right, John, what A.I. Question to be afterward. So I think we talked about this before the show. We talked about software development, the history there, very deterministic. And now, and then we started talking about data and how data and data teams are less deterministic, right? They're dealing with fuzzier problems and fuzzier outcomes. And then AI comes in and you've got a bunch of like historical deterministic technology teams being said, hey, implement this A.I. thing, right? And I think it's safe to say it's often not going well. So my question to you is, A, and ensure there's probably going to be some changes on, you know, on the traditional, you know, deterministic side anyways and like traditional I.T. My question to you is it's our data teams may be better suited for some of these A.I. Implementation A.I. problem solving because they're used to the less deterministic working style. Yeah, I do think so. For me, data teams should be problem solvers, first and foremost. Whereas I know a lot of software engineers will say they are too. But the way I see the teams work are very different. A good example. So when I was at Best Buy, we were asked to be part of the beta test for a new version of Power BI. And they brought in IT, they brought in my team. We're sitting there asking questions left and right. And IT is just like, next step. OK, next step. I'm like, no, why don't you guys ask him questions? That to me was the difference in the mindset right there. It's for us when we're given a problem. It's, hey, is this even the right problem we're solving? It was this frame. That's where we start out with an scrum process, can band. They're great at a certain point, but in the beginning, there's really a big difference between software engineering and data and analytics. And when it comes to the data engineering side, you see some people like, oh, well, data engineers should really be an IT and it's like, well, depends on the org. But if you're solving problems on the analytics side, you need those data engineers sitting day to day with your analysts and your data scientists. And when it comes to AI, like I mentioned before, I do not use open AI or co-pilot or Gemini. I experiment with the cloud, but the way I use cloud is really conversational. Yeah, I find a lot of people when they start using AI, there's like, do this. Now do that. I we have an ongoing conversation on various threads. Because to me, that's the best way to use it. And I think people that work in data and analytics, they're used to that asking questions, one, the answer, having that conversation going back and forth to find the answer. So that mindset, I think works quite well if you use AI correctly. Yeah, yeah, I think that makes sense. But since AI is becoming so prevalent, wouldn't you think that maybe that deterministic mindset, at least to an extreme is not going to work for anyone, like long term? Yeah, but at the same time, it's so entrenched. It's going to go away. Yeah. And there will be years and years of like work to do and things for people to do of systems that are deterministic and need people to work on them. And so that's going away overnight. And I think there are fields, you know, like you want something deterministic when it comes to like your finance management, you know, you're in healthcare, so they're always going to meet that space. And like, you know, when I hear people saying, Oh, we're going to use like, you know, AI to be like a replacement for nurses. I'm like, yeah, I hope I never end up in that hospital. Yeah, yeah, yeah. I mean, I use it and I'm just like, if I use it for things that I'm familiar with, that I just want to help me speed things up so that I understand what the output is. But if it's something I'm not familiar with, how do I know to call it out when it's blocked, right? That's what I see people doing or they just get lazy and they don't really look at the responses. You know, like I've used it. Yeah, when it came out and I was using it for like, you know, writing cover letters and, Hey, make my resume adapted to this job. And you know, I ended up with like PhDs from Stanford working at like Facebook or Google, which I never have. Wow. I'm like, no, I don't want you to make stuff off. I want to be fast, like use the key words. It's in the job descriptions. Right. Right. Yeah, super interesting. Well, I think we're at the buzzer here. But Edward, one more question for you. You mentioned that you've been rereading some of your data books, you know, from a decade plus that are a decade plus old. Do you have a couple book recommendations for the audience on the ones that you know, you return to most often? I can tell you the one I'm reading, right? Rereading again, the connected company. Hmm. Very interested one. I'm like, it's by Dave Gray. I like the book because he uses a lot of different concepts that. So I believe I read this back in 2012 or 13 concept. So I'm like, they're still relevant today. I would see here. Another one I liked. I actually met this guy at the same time. The Intention Economy by Doc Sirls. Hmm. So I read this one over the summer. Again, if you're looking at like personalization, I think it's a great book. There's a companion book that's more technical called the live web written by professor down at the University of Utah. But it's a great book for like, hey, how do we actually create an economy that's much more driven by the consumer? And I think it's actually very relevant today, whereas, you know, we see companies that are really shareholder driven where, you know, they've got. They'd record profits, but it wasn't good enough. So they lay a bunch of people off. And I'm like, that's not sustainable guys. Right. You know, and I've been having these conversations with a lot of people. It's like, Hey, the younger generation in their twenties, they don't want to be doing like what we're doing 20 years from now. They want to, they want to work environment that's much more for them. And much more, you know, satisfying in terms of giving them a rich life. It's not all about like, Hey, I'm just here. So some shareholder can make money. Yeah, I think it's a great book talking about like, Hey, how could you do that? Actually, I've been talking to a founder, a guy. We actually went to the same school together, Thunderbird. And he has started a company, a bronze. They started a company called hiring, hire humans. We're putting a lot of those concepts into place about how do you improve the hiring process? So that's something we actually were stalking this morning. Those are the kind of companies I'm looking forward to see, like becoming coming out there and driving things because I, you know, things change. And I'm looking at those things up like, Hey, how do we adapt to, you know, the younger generation and what they're looking for, not even the younger generation. I mean, I want to work for the mode. I'd like to go work at a ski resort on my computer. I don't want to have to sit the office all the time. So I think it's just new mindsets coming in. So I'm looking, you know, those books I look for, these other books. Let's see here. I don't have it with me, but some of the books on just like, how do you engage with customers in different ways? How do you look at? How do you look at the perceptions people have about what it is you're creating? So I, like I said, that whole product mindset I do bring to the table. So looking at things like, you know, going back to this conversation on hire humans, it's like, Hey, the job is a product. The person's a product in some way. So how do you do? How do you make that work better? Ben and I were just talking about, it's like, well, most people don't know how to actually hire people in air. True. That's the process of what's broken. So that's what I'm, how I spend my time, what I look at. What I've just found is like these older books just give me the details and information I need more than the stuff that's coming out today. Love it. Well, we'll try to put those in the show notes for this show. And in the upcoming newsletter, Edward, this has been an amazing conversation. What an incredible journey that you've had and excited to see where you land next. Once you get in there and, you know, and start causing trouble like you did at all these other companies, I would love to have you back on and hear about it. Sounds good. Thank you for having me. The data stack show is brought to you by Rutterstack, the warehouse native customer data platform. Rutterstack has purpose built to help data teams turn customer data into a competitive advantage. Learn more at Rutterstack.com. [MUSIC]
