Delegate to Innovate: How Letting Go Makes You a Better Leader

0:00:05.2 Satyen Sangani: Welcome back to Data Radicals. Today's guest has had a career like no other, from shipboard operations in the Coast Guard to leading AI at one of America's largest grocery chains. I'm talking about Todd James, former Chief Data and Technology Officer at 84.51, the data and science arm of Kroger. Todd shares how he helped Kroger use AI to speed up grocery delivery and boost customer experience, all at massive scale. We also talk about what makes data teams work, why the CDO role may disappear, and how the best leaders know when to step back and let others step up. If you're building AI, leading teams, or trying to turn data into impact, this one's for you.

0:00:47.5 Speaker 2: This podcast is brought to you by Alation, a platform that delivers trusted data. AI creators know you can't have trusted AI without trusted data. Today our customers use Alation to build game-changing AI solutions that streamline productivity and improve the customer experience. Learn more about Alation at alation.com.

0:01:09.4 Satyen Sangani: Today on Data Radicals, I'm excited to be joined by Todd James, the founder and CEO of Aurora Insights. Todd is the former Chief Data and Technology Officer of 84.51°, a retail data science, insights and media company and subsidiary of Kroger. Previously, Todd spent 15 years at Fidelity where he built global data and analytics capabilities and led AI driven modernization. And before that he held leadership roles at Deloitte and served as an officer in the US Coast Guard where he led initiatives in IT, information security and shipboard operations. Todd, welcome to Data Radicals.

0:01:45.1 Todd James: Thank you for having me.

Todd’s career journey: From the Coast Guard to Fidelity

0:01:46.8 Satyen Sangani: So you have a pretty unique career journey and also a pretty unique recent title in being both a data officer but also a technology officer. Tell us how you sort of got down that path and why you mixed both data and technology in a way that few leaders seem to have done.

0:02:03.6 Todd James: My career has always been rooted somewhat in technology. I've either been on the business side with a heavy focus on how to use technology to advance the business, or I've been on the technology side, leaning towards how do you have a positive impact on the business? So I've always kind of straddled that, and I found a lot of value throughout my career in stepping away from technology from time to time to serve in business roles to give me that other perspective; what's it mean to hold a P&L? What's it mean to run a large global operations? So that's been part of the journey.

From a technology and data perspective, at the end of the day, my passion is probably more around the activation of data through advanced analytics and the impact that it can have on the business. I found myself in the data roles 'cause you can't do a whole lot of activations without the ability to get to the data. So that's always been a natural flow to it. The technology is just an extension of that. I think the power comes when you're able to put technology with the data, with the analytics, in a way that serves business objectives.

0:03:04.7 Todd James: And so I actually found that role kind of flowed together very well. Also, keep in mind, I was the chief technology officer at a data science and analytics company, too. There's probably a little bit more natural flow between those roles at a data firm and the information firm than, say, at a major retailer. So there was some difference there.

0:03:24.2 Satyen Sangani: Yeah, no, 84.51° is a pretty unique place, and we should get to what that is. But as you think about sort of your history and your career arc and trajectory, I mean, you started in CS. You then move on to, obviously, the military and the Coast Guard, and even being involved in a shipboard operations is a pretty physical thing. You then move on to Deloitte and do some consulting, and you then do Fidelity, which is obviously financial services, and now here you are in retail. Those are a lot of different domains. What do you think it is that allowed you to succeed in all of those sort of disparate places? And what are the skills that you come back to that sort of says, this is why I'm where I am? 

0:04:04.2 Todd James: It's interesting. I had a similar question. I was talking to a group of master's students at the University of Cincinnati a couple of weeks ago, and they're asking for my guidance on where you should go, which it's quite a privilege to be able to have those discussions, and what you should focus on from a career perspective. 

I think that the most important thing that has helped me, and I can tell you where it started, was the ability to be agile, to be flexible, and to be a voracious learner. I remember when I was leaving Deloitte, one of the partners, as I was walking out the door, had said, you know, Todd's a sponge. And I went, I'm like, what's a sponge? Why is he saying that about me? I mean, is that good? Is that bad? And then he went on and he mentioned he soaks up information and figures out how to apply it. 

At an early stage in my career, one of the things that most people may not know but like within the Coast Guard, you actually have to do a lot of different missions. You could be doing law enforcement in the morning, you could be doing marine environmental protection in the afternoon, you could be responding to a search and rescue call later in the evening.

0:05:04.9 Todd James: So there's a lot of context switching, and you have to learn quickly within each situation: what do you need to know about that specific situation, how do you apply what you've learned. And so there's a lot of diversity that I think as I was coming, I went in the military at 18, I came out just around the time I was 30. So at a formative stage in my life, there was a big focus on having to switch context. Even within the service, you rotate out every two to four years into a new job, from being on a ship and running operations to going to grad school to running a big component of their IT. And so there's a lot of context switching there. And that stuck with me throughout my career. I think to some extent it energizes me. I enjoy the learning. So it's somewhat self-reinforcing. You're good at it 'cause you enjoy doing it. But consulting further reinforced that kind of context switching as you move across industries and you sell different services. And Fidelity was another one of those companies that I went in and I got in the door and said, I want to do tech, I want to learn financial services.

0:06:09.4 Todd James: So I focused on areas of the business that really didn't have anything to do to tech. It was only in my last couple of roles there where I really started to get regrounded in Artificial Intelligence in the journey and get back into the tech side of an organization.

0:06:22.2 Todd James: So it's I look at it, that's how I manage my entire career. I think the root of it is, you got to be a lifelong learner. You have to be very flexible, you have to be very agile. I also think for, as you look at a career going forward, that was important across the span of my career, I think it's going to be even more important for people as we look out over the next 10 to 20 years and what the careers for people that are sitting in college today are going to look like and how those are going to change. 

And I'll even go as far to say, and you could probably relate to this when I came out of college out of my undergrad computer science degree, the job that I just left at Kroger in 84.51°, that title didn't exist.

0:07:02.1 Todd James: Wasn't a job. So to some extent we talk about the agility, a lot of what we prepare ourselves for early on in our career, and throughout our caree,r some of the roles that we thought would have don't exist and a lot of the roles that we end up taking didn't exist at that time.

0:07:17.1 Satyen Sangani: Yeah. The thing about context switching, and it's funny because we, I was talking about this with my head of engineering this morning, is that if you give people too much context or too many different threads to pull on, often it's really hard to actually execute because you're sort of running in so many different directions. How have you balanced that for yourself, and how have you gotten to the place where it's not just about like learning about this miasma of stuff, but actually focusing on a couple of things that matter? 

0:07:45.4 Todd James: I think one, you have to have enough wisdom to know how big a swing to take, right? I think the old adage is: you can change industries or you can change functions. You don't want to do both at the same time. So there are some rules that it suggests that people follow that I've tried to follow throughout my career. 

You said something really important there too, that I want to reinforce, no matter what you're doing, an ability to focus and prioritize on what's most critical, what's most important. Early in my career, and I don't know where it came, but I envisioned it was probably starting and you get all this work loading up. And I envisioned that my job was a big plate of spaghetti that big one, it's overflowing over the plate. And for me, it became as important not to figure out what meatballs I want to keep on the plate, but what ones I want to let roll off. And I know it's a stupid analogy, but as I get into a week, I try and figure out what are the things that really don't matter that I don't, is active is what I need to focus on? 

0:08:44.5 Todd James: What are those things that may be important, but if I can't get to them are the right things to let go? And you do that strategically as you look out over a year and what you're going to prioritize and what you're going to do in a role. You also do it tactically on a day-to-day basis, try and get through it. 

And I think if you're not creating some space to have those decisions, like my worst days are the days I show up and I have not had the opportunity to go through that calculus, 'cause when that happens, you tend to get overcome by the day. And if it works out, you're lucky. It's certainly not because you were planful about it and conscious about where you want to put your time. And I think we all have days like that, so I try to minimize those days.

0:09:26.4 Satyen Sangani: Yeah, it's a really interesting skill for data leaders in particular, because, I mean, you're constantly in this place where you want to organize everything. You have a lot of data coming at you. You're probably in the job because you like things being properly categorized and neat, and yet you also have to go deliver. And it sounds like some of the habits that you have are sort of really just to sort of take the space back and sort of saying, look, what is on my to-do list? And in your analogy, like, okay, these meatballs just might actually make me bloated, so I'm going to actually just put them off my plate. When did you start doing that? And what other practices do you have to sort of help you get centered? 

Great leaders delegate: Advice for aspiring executives

0:10:03.1 Todd James: Yeah, I think part of doing that, I think there's a point in your career as you're coming up. And I remember the moment it happened for me. I was at Deloitte, and I had taken over a very large account running strategy for Homeland Security for the company, the business in TSA. And I still had some commercial work outside. And I was having dinner one night with a gentleman who was a very senior director in the firm, towards the end of his career. And I had the privilege of, he and I were both Geo Bachelors in D.C at the time, where we were spending a lot of our time. So we'd do a lot of dinners, and I would learn a lot from this gentleman. His name was Dennis. And one day he looked at me, probably 'cause he saw something he didn't like. And we're sitting there over a steak dinner, and Dennis looked at me and said, how do you know when to delegate, Todd? How do you know when to kind of prioritize and give things away? And I'm like, well, that's easy, Dennis. Like, if it's not going to get messed up, I can give it away.

0:10:55.9 Todd James: And he's like, that's wrong. And that's why you're killing yourself and you're getting mired down in the work. And I said, well, what do you mean, Dennis? And he said, you delegate everything that if it breaks, and you can put it together, even if it's painful, those are opportunities for your team. Those are opportunities for people to grow. You may take a few hits, but if you can fix it, why wouldn't you delegate it? The things that you can't fix, the things that are super high risk. Yeah, you got to stay more involved in those. But I thought it was one humbling to be sitting there at dinner thinking I'm doing pretty well. I got this big account, I'm having a big influence in a new space. And it was a little bit humbling, but it was also very insightful to hear that. And I gave that advice to a couple of people at Kroger just a couple of weeks ago when I was having a discussion with some, and they were kind of at that stage in their career and working through it. 

So I think a big part of it for me, and I see it in a lot of people, you get to a point just as you start to escape that velocity, where you're the best person in a particular technical discipline or field and you've really got your arms around everything, you understand the detail to where you get rewarded.

0:12:09.6 Todd James: And I'll put that in quotes. Right. "Rewarded" to be able to take on even broader scope. That happens to all of us at some point in our career. And I think that's, that was probably the most formative point at that time, where I was really taking on larger and larger groups distributed across geographies, different teams, high-profile work happening in these clients, where I realized that I had to change how I was leading. And it was one of those times in my career where I look back and say this was an incremental improvement. I came out of that job with a step function change in my abilities as an executive and my abilities as a leader. And like so much in life, sometimes it takes someone sitting across the table and figuratively slapping you in the face with much better wisdom than you brought into the room.

AI retail use cases: Quicker routes and personalization at scale

0:12:53.5 Satyen Sangani: Yeah, it's a really helpful and interesting framework. So the 2 by 2 is: can't fail and if it does fail, can't be fixed. So that's super, super helpful. 

Your most recent role is at 84.51°. Before we get into the role and the work that you did, which is super interesting, maybe you could describe for our listeners what is Kroger and then within Kroger, what is 84.51° and then within that what role you took on? 

0:13:20.4 Todd James: Yeah, I'd be happy to. So Kroger is a Fortune 25 grocery retail company. It's the largest traditional grocery retailer in the United States, covers the majority of the geographies across the country, but basically largest grocery store in the United States. Within that is a data science company called 84.51°. And what 84.51° does is a few things. One, it's a data science company, it's a retail media company, it's an insights company. And how does that manifest? It basically takes data from the Kroger Corporation and uses it to provide greater insights on customers path to purchase which helps CPGs better understand consumer packaged good companies better understand what's happening with their products inside our stores. Then we have a retail media company that, using that data gives them an opportunity to activate against that information, to be able to provide promotions, and to be able to think about how they may want to position their product differently within our space. We also do some other things. I mean we have a venture capital arm that uses data science to look at food services businesses and to be able to figure out where you invest. And when we do invest, a lot of times we'll bring them in, put them on Kroger shelves.

0:14:37.8 Todd James: And the other big component of it is the in-house data science team for the parent company for Kroger. So within 84.51°, I was the chief data and technology officer. So I was running the data and technology for the 84.51° business. About 60% of that was commercial-focused businesses. About 40% of that was focused on helping the Kroger company change their business models through data and advanced analytics. 

The role was also kind of interesting in that I had two roles as Chief Data and Technology Officer at 84.51°. But I also sat on the Kroger technology leadership team. I was an executive within the Kroger company, where I was accountable for the AI strategy and enterprise data as well. So what was neat about the role was it was a role that gave me an opportunity to work across two very different companies, right. You've got a small retail insights business that operates and moves at one pace, then you have $140 billion company that has different challenges and different problem sets and you have to think differently about the problem sets in each area. So I mean it was a very exciting place to be, and then figuring out how you pull it back together.

0:15:49.9 Todd James: I thoroughly enjoyed my time at Kroger and 84.51°. It was just a fantastic place to be.

0:15:55.0 Satyen Sangani: Yeah. And tell us a little bit about the recent work that you did there, because I mean, if you think about it, obviously, grocery chains, super low margin, obviously, massively high volume, efficiency is everything. And then within that, you've got 84.51°, which is really sort of in some sense becoming the brains of the operation. And so, how to route things like all of the work that gets done in order to be able to figure out what to sell, where to sell, who to sell to. Large part of that's getting done there. As AI became more prevalent in the last few years, what were the projects that you worked on and what were some of the things that were top of mind? 

0:16:35.4 Todd James: It transitioned over time and we can talk about these later. I view chief data and analytics officers as a transitional role. A moment in time as we go through right now, is what I would call a once-in-a-lifetime re-platforming that we're going through. 

So my initial focus coming in was really to work, I probably spent about the first year working within 84.51° helping them think through how do we change from being a data consulting company. Grew out of consulting, had started as you know with Dunnhumby and then an acquisition that brought a component of that business into Kroger, which became 84.51°. So it was data science and consultants, power of two coming together to be able to generate significant insights. 

What we started to realize is more and more: that model doesn't work well when you want to operate at scale. And there's some limitations to pure play consulting. That industry-leading model. When they started it, there was an understanding that we needed to focus more towards tech services. So about my first year was really focused within 84.51° and working on how do we become more of a tech oriented business, how do we think differently about how we organize, how we think differently about the capabilities we need and the focus that we're putting on the commercial side as well as what the future Kroger model teams look like.

0:17:53.5 Todd James: After that I got more and more involved with the parent company. And when you look at the legacy of 84.51°, it comes out of that marketing and merchandising. So when I showed up, there was a heavy penetration of data science within the merchandising and marketing space within Kroger, which makes sense based on if you think of Dunnhumby and where 84.51° came from, that was a very natural and what I saw there was, I mean there was a lot of maturity with capabilities and platforms that existed from a science perspective in that space. 

Now we did more, and we'll talk about that in a bit. But the other broader area, there hadn't been a lot of penetration across other areas of the business. So one of my big focuses was how do we get support from the broad leadership team to understand what AI could do? I had good partner in the CEO. His leadership team spent a lot of time with them early on saying this is how we need to think differently about AI in the future. AI should not be something that I'm going to the board talking about, but it's something that Kroger leaders should be embedding in their strategies.

0:19:05.0 Todd James: And I had great partnership with the Kroger executive team to do that to a point where the strategies that they had started to embed AI end to end. And as we did that, we were able to walk the impact of artificial intelligence more broadly across the entire business, which was good.

0:19:23.9 Todd James: Which led to a different challenge of how do you do it at scale? Which is something that has been the big focus of last year. 

But when you talk about the types of initiatives, a few things come and pop. One was in the work that we were doing around personalization. There's always opportunity to improve the sciences to get better returns. But there was also a big focus on how do we get more real-time science as opposed to handing scored models over the fence from 84.51° at Kroger. So there was a big focus early on in enabling real-time, point-of-action sciences into the customer journeys that were happening in Kroger. So kind of a shift on that. 

The big areas. It was interesting when I was looking at the numbers just a couple weeks ago, but like the retail operations and supply chain spaces were relatively underserved.

0:20:11.2 Todd James: There was an opportunity, and I said let's build full-stack teams. We know that those are areas where we want to deprioritize just providing insights as questions come in, being that question and answer queue. And what we want to be able to do is go after activated sciences that actually interact directly with associate journeys. 

So we focused early on, on some capabilities around the E-commerce experience. So one example I would give you is looking at something we called dynamic batching and routing.

0:20:43.0 Todd James: And the way to think about it is if you get on your Kroger app, which anyone listening to this should maybe you're hungry great company. But get on your Kroger app and you make an order. If you're going to do a pickup order at store, there's actually a store associate that goes around and picks those items. And if you think that Kroger has 2500 stores, the challenge was, how do we look at being able to have an impact that can actually enable us to reduce the lead time that a customer has for making that order? And the idea was, is if we could route the associate around the store quicker and more efficiently, it'd be better for the associate, it'd be better for the customer.

0:21:21.2 Todd James: And so we were able to put in advanced routing algorithms that reduce the distance traveled on a pick order by about 10%. And what that enabled us to do was to pull back the lead time, better customer experience, better associate experience, 'cause they were doing less hunting and picking, they were more efficient in their role. And so, really good outcome for the associate experience. 

And the beauty of grocery retail. When you have that kind of platform, small changes have an enormous impact. So there was, I mean, there was tremendous benefit to being able to drive that across all of our stores. And we were able to roll that out in around six months. I mean, it was a pretty aggressive rollout considering the complexity of the science working to change processes in store. That's a big part of it. You've got to not just put science in, but work changes. And so you're working with the associates. We embedded data scientists out in the field to make that more successful. But one of the things we also did is I said, I think we're going to use this again. It's a routing algorithm, right? And everyone's kind of agreeing and saying, yeah, we're seeing the same thing.

0:22:20.7 Todd James: So there was a lot of focus on hardening it and positioning it for scale. And after we got the dynamic batching out, we said, well, if we can route associates around the store, we could probably route trucks from a distribution center to our stores. So we immediately took that algorithm instead of taking. Took us about nine months to build and six months to roll out, within three months we were rolling this out one of the facilities in Texas for supply chain distribution center to store routing. And the target was we thought we could get about a 3% reduction in distance traveled, which is good for cost, it's good for flexibility, and it's good for, like carbon emissions. Last test I saw on that, and it was just a few weeks ago, it was about 8.63%, which was a really good reductions. So we were able to drive through scalable sciences, not just real-time, point of sale having the impact on operations, but the other mindset that came through, how we started building sciences in these places is we have to start to think about building a layer of algorithmic infrastructure that allows for reuse.

0:23:25.7 Todd James: That's how we're going to get speed. That's how we're going to be able to manage at scale. So that was a big part of the journey that we were on. I mean, otherwise doing great things with the marketing group at Kroger around being able to generate advertisements. You've got the personalization, substitution sciences, advanced price and promotion. There's a lot of areas where you're touching basically every aspect of a retail environment. But I would say when I look, the big focus that I had is how do we fully penetrate the space to have a more holistic effect across the entire company? And how do we start to position to be able to operate at scale? 'cause ultimately the desire for every company should be, I want to positively impact every single decision point with data and advanced analytics, either through informing to make a better decision, like a leadership decision, or activating where you can into process flows and experiences.

0:24:21.7 Satyen Sangani: Those are two phenomenal examples. And maybe starting with the first one, how did you decide that this traveling, like, how did you first recognize that problem and where did sort of the problem statement originate? 

0:24:32.9 Todd James: One of the things that I think is really important is to get close to where the operations are happening. So back when I said we need to have a broader penetration across the Kroger company, getting into areas that were underserved, the first step was to go out into the stores. I mean, you're going out with the leaders from the retail operations group, you're working closely with the supply chain. But it's not just meetings. It's going out into the store and doing some of the work. 

I went out early on, learned two things. One, it's a great experience to be able to do some of these jobs. Two, I also realized after I did one of the jobs that our science was helping someone with. They had to redo it. I wasn't very good on it. So I knew that if the data science stuff didn't work out, the store was not going to hire me. But we've got the data scientists, the engineers, the product managers out in the stores and talking in generating ideas with the leaders. A lot of times these ideas were floated back and forth in a Kroger store, not necessarily in an office.

0:25:33.7 Todd James: Everything would go back and you get in front of the whiteboards eventually. But it was about getting really intimate with the business. And there's two things going on. One, the business is understanding what the capability from data science can be for them. So they're raising ideas as they understand it. On the flip side, the data science and technical teams are understanding what the needs of the business are. Now ultimately there was a prioritization with the senior leadership within retail operations, but the ideas tended to come more organically through collaborations at multiple levels. Is where I found we came up with some of our best ideas. Now...

Reapplying data-science processes across distinct problems

0:26:12.3 Satyen Sangani: Yeah, that's sort of the classic, classic entrepreneurial advice of like get out of the building. In this case this would be like HQ. And in your case, getting into the field where your end customers are. The second problem that you mentioned, the sort of truck side of the distribution center problem, actually feels characteristically different because you've got infrastructure like Google Maps and traffic where you would think, well, a truck can only travel really on one road at a time. That feels like a slightly different problem. Tell me about what's overlapping and what parts of the algorithmic you reuse. Like what parts of the sort of models did you reuse in one versus the other? And that sort of infrastructure of algorithms and models that you created, like tell us a little bit about that 'cause that seems pretty interesting. And I think a lot of people talk about trying to do that, but struggle with actually doing it? 

0:26:57.7 Todd James: Yeah, I think you're raising a very fair point. It's not like cookie cutter. Hey, it worked here, and I put it here and magic. And so yeah, I don't want to imply that we looked at three angles. What were the engineering patterns that we used for really robust real time sciences? I mean those store sciences are doing 200,000 route combinations a second. That is very critical, complex, high-end science that requires a lot of engineering behind it to make it work. So what engineering patterns could we use and could we apply? And so we were able to take some of those that we thought were relevant. 

The other part we don't talk about a lot is what about process? We started putting some of these real time sciences and it kind of broke our existing support models. How are you going to start to support sciences that are actually physically running operations in a store and what happens when they go down and how do you think different about that? So there was a process that poured it over and part of that process was a process I described with you. We weren't going to do anything without going out and getting intimate with the people that actually do the work.

0:28:03.7 Todd James: That was one of the processes we figured out that worked. We want to do it again, but also the back-end support processes. And we started to think, well, let's not do it separate. Let's build out what we need for the store Ops. But then let's also say whatever we're going to do for the supply chain component as much as practical, we want to be able to build a common support structure for real-time sciences at scale. So that helped in the past. 

To be honest, each product team supported their own and we realized that you get to a point, and I learned this at Fidelity, you get to a point where that works for a while, it's actually advantageous. But once you start to get 60% of your time on support and only 40% on new development, you've gone too far. So you want to avoid that. On the science, the volume, the data very different. You're still liquidating across all possible paths and different combinations to be able to opt it still goes back. They're both the nature is a traveling salesman problem that we were dealing with. So I would say how we're looking at the we use new data, right.

0:29:09.0 Todd James: And we went through but it's the same nature of problem which meant that we could share a lot of knowledge and expertise from learnings from what we did in retail Ops with the supply chain aspect. But yeah, you're doing on different data sets. So there, there are some differences there.

0:29:24.7 Satyen Sangani: So these are more foundational and procedural and sort of ways of working oriented bits of reuse than they are like oh, I've actually got this specific algorithm and I'm actually going to reuse it in this other place. It's really like how do we recognize the patterns and the problems and the people and the processes so that we're using the same things, that we're creating the same sort of patterns of reuse over and over again? 

0:29:44.6 Todd James: Yeah, I would put a little bit differently and did the same thing at Fidelity. We had entity extraction work that we used for being able to help B2B client service staff in different areas of the business be able to rapidly extract information to reduce the, I think in financial services and the data is a little bit stale, but I think at the time 30% of an associate's time was looking for stuff in these client services roles. So being able to reduce that through semantic sciences and do data extraction, what I found there and what I think is the same you have underlying principles, you have underlying approaches, you do put different veneers on top a little bit in terms of how you expose it. There's more commonality I think than what you apply to. And I think that it's manifest to the fact that in the routing algorithm example, what took us about nine months to build, took us about three months to get ready for our first run with a distribution center down in Texas. So could you say it's procedural, algorithmic? I'd say we ported all the algorithmic learnings, but yeah, of course you're running some new sciences on new data.

0:30:55.2 Todd James: I would say it would have taken a long time had we not had the structures in place to apply those learnings. And the other reason I'm a little hesitant to play it down now that I'm no longer at Kroger, it's up to the resources there. But we were talking in terms of we're building a routing center of excellence 'cause we can take that. The idea being after distribution to stores, you can take it to a lot of other aspects across the organization where you're routing. And our expectation would be that each subsequent application of that capability would take lower unit times and lower unit costs to be able to implement. And going back to the examples from financial services, as we walked against the different client support teams, that's exactly what we saw the first time we implemented the data extraction.

0:31:42.6 Todd James: And this is before ChatGPT was out there. This was hard to do back then. We found reduced unit costs and reduced times to deployment across the board. And that's why I called it kind of a veneer. You got to put a little bit of the flavor on it to make it relevant for this client service group that's focused on mid market versus this one that's focused on tax exempt.

0:32:03.1 Todd James: And I think same approach that we'll see with the routing examples across retail.

0:32:07.7 Satyen Sangani: Yeah, makes total sense. Those are fabulous examples of what worked, what didn't work that you thought might work or where did you find that either AI application or maybe it was the way the teams were structured, like what failed 'cause you learned so much from the stuff that didn't pan out relative to the stuff that did? 

0:32:27.4 Todd James: Yeah, I think one of the biggest challenges, and I think across probably two flavors of challenge, one is when you're working with a group and you haven't done sufficient work with the people that are going to be impacted to get them over, to get not only their buy-in, but I think in some points, like a legitimate fear of what this science is going to do for them. So I can think in both companies where there was a technical solution, leadership was on board, but we hadn't invested enough time in the resources that were actually being impacted. And I tell you what, if the people at the consumption end of your sciences aren't bought in, they don't get implemented. Yeah, I didn't do it. I didn't use it, wasn't able to do it. And that's where I found we had working sciences that were performant and didn't get adopted. The other example that I would say is it's probably a different flavor of the same problem. There are a few examples where we had multiple constituencies looking for a science to do something, and we got to a point where we would have a one or two things would happen.

0:33:48.1 Todd James: One, you would have a science that would work for one group, not to the other, but you needed everyone on board so that wouldn't get adopted. So kind of the stakeholder management and being very clear and being able to get them. These things didn't happen a lot, but you can tell from my tone, they're learning moments. The other area that I would say too is I remember a case where we did have a science that was able to go out and able to look at relationships between variables at a store level that impacted revenue. And I mean, the science was really good and it was able to tell store by store the relative impact of these variables. And it can be different from store A and store B a few miles away. It wasn't like a generic model. It would look at the data at each individual store and give us that information. The challenge is we'd be able to identify it, but the prescriptive aspects weren't sufficient to take action.

0:34:49.1 Todd James: Like sit here and say, well, if you know this, can't you get there? Can't you look at it? But you've got to think within a large organization who ultimately is going to consume the science? And is that science able to provide enough specificity, not just around the problem, but about the action to solve it, to make it something that was significantly better than what they were doing without the science?

0:35:11.4 Todd James: And that's always hard when you have a really good science, but you're just not quite there in the ability to solve the problem. And those are the tough ones to kill. I think the biggest mistake you make with those is not killing them. And that one we ultimately said we got other things to work on. This isn't going to hit. We'll document it, we'll pull it up later. But after a year pushing it uphill, you realize that you're not close enough to being able to have a science that's going to be able to meet their needs and therefore you have to back away from it.

0:35:42.2 Satyen Sangani: Yeah, the notion that sort of says, look like we need a willing customer and we need to give them something that actually makes their life better. Seems like very obvious things to say. Like when I say, when you say it, like the way I said it, like it seems like very obvious, but like then you're in the midst of this big organization, you have these great ideas, everyone wants to do AI of course, it's like a great idea. If you do it, of course they're going to use it. And then of course, if you do it, of course it's going to be relevant. But then of course those are not things that happen.

0:36:11.2 Todd James: You know what though? I look at it and I bring it up 'cause I think that is something that whether you're working with a company as a vendor or you're a CDO within a company, those are learning moments. Those are things that didn't go well. Those are things that I actually look back on and say my teams and I, it's kind of like a marriage, right? You need both parties to come together. But there's more. I look back, I think the accountability. There is accountability on the CDOs and their teams. There is accountability on the vendors for working through that. And that's often the hard work that I think gets overlooked a lot of time. So we took those learnings, we applied them, spent a lot more time in the stores, a lot more time working out. But I don't think that's something we have to accept. I think we should go into every project, every initiative, saying I own an outcome around, bringing people along and convincing them. And if you're doing that right, you're probably spending more than half of the project or half the initiative on managing those organizational dynamics, working with people and how they think and how they feel to be able to drive them to an outcome, to listen.

0:37:19.6 Todd James: I think that's more than half the work that needs to happen in the spaces. We talked about data and analytics and how hard the math is and how cool the outcomes are, but this is transformation, this is about people.

0:37:32.0 Satyen Sangani: Yeah, I think that's the key observation that I think most analytics leaders, you want to really believe, well, it's going to be better if we build it, they're going to come. And of course they don't. And it's because either they don't understand it, they're not bought into the change, they don't actually think it's a real problem. But you don't really know any of those things unless you get out of the building. And the great part of how you've narrated every single one of these problems is that it's just literally started with a business problem and it's worked backwards as opposed to, oh, we're going to go build this thing with Databricks or AWS or Azure, and then we're going to go install Alation and then people are going to do some discovery and then they're going to go find some data. And then when they find some data, they're going to actually sell some. It's like, wait a second. Like nothing in that narrative has a business problem anywhere in the zip code of being solved. And you've really just worked, like consistently just talked about it, thought about it, and discussed everything from the problem statement backwards.

0:38:27.5 Todd James: First of all to, those tools are helpful. So all those tools you mentioned are helpful. We need those tools. I think for me, having spent about half my career in the business and about half in tech, the orientation is to start with the business. And the other thing too, and I learned this early on in my last company where we spent a lot of time trying to build right data lake before we started analytics out of tech. And then we said, hey, this isn't working, it's taking too long. Let's line up all the use cases, figure out where there's critical mass around data, and we'll move the data as we do the use case. And that puts a little bit of a higher tax on your upfront use cases.

0:39:08.4 Todd James: But we're able to drive value as we push down the funnel. Even though there's some inherent inefficiencies to that. I don't know how else you do it. I talk to my peers at some of these other large companies and you'll see some that have really put in a beautiful foundation layer. And when you talk, some are seeing success, but a lot are saying I can't get people to use it.

0:39:30.5 Todd James: I would much rather be a little bit inefficient, create a little bit of throwaway, but start driving business value today to fund my investments in the foundation as I go along. And it also gives you the opportunity to understand what foundation is valuable. 'Cause if you and I were to sit here for the next, we could come up with brilliant ideas on what are the foundational elements for a particular company based on their needs and what they need from AI. We wouldn't be right on all the points. So being able to build out as you're driving value and learn along the way, I find that to be a better approach. And it gives you an opportunity too. Everyone says you got to wait for all the data to be in order. Well, when the data's in order, things are easy. But the only way the data gets in order is if you make it in order. And the only way you get funding to make it in order is if you tell someone you're going to deliver on value.

Is the CDO role transitional?

0:40:21.5 Satyen Sangani: Yeah, that's exactly right. You have to work in the right order. And the order of things is solve a business problem and then we'll give you the ability to sort of fix the data to get there, not fix the data. So you could solve a whole bunch of random business problems that nobody's asked for. 

You mentioned something, I think really I guess, I don't know. It seemed like a pretty provocative statement that the CD, CDTO or CTDO, whatever that this job is transitional, say more? 

0:40:50.2 Todd James: I don't think this job lasts forever if properly managed, there will be a title CDO or it may be headed data. But here's the trend that I think is going to play out and I'll tell you why I believe it's going to play out. Right now the coolest thing about this chief data and advanced analytics role is the analytics side, the activation of data. You do the data side so that you can activate against analytics to drive transformative value. That's what gets people excited in these roles. That's what gets me excited at least. But where I think this trend is going to head similar to other evolutions across industry. You can go back and you can look at the lean Six Sigma and new capabilities that came out. At some point there is going to be enough knowledge around the application of AI in the business that a lot of the really exciting aspects of deploying advanced analytics are going to be part of the business roles, in my opinion. Then I saw this when I was at Fidelity. Like we would have some of my AI product managers, they would get intimate with the business, they'd move in and start running a business unit.

0:42:02.3 Todd James: What would we do after they got there? We'd push the advanced analytics resources closer to them 'cause they knew what they were doing with them and it got even more embedded within that space. So I do think we're going to see that trend. 

The other trend I'm starting to see, that kind of supports that is coming out of business schools I am seeing. We recently hired someone who was undergrad in supply chain, went to work for one of the large tech companies in their supply chain practice, went to a good technical school in the Northeast and came out with a data science degree. And so supply chain and data science across two degrees. One, to work for us. And what I think is interesting, I don't think she wants my job or what the job I had at the time. I think she wants to be head of supply chain. And so she is kind of building that technical. And I've seen that in more of these business analytics programs. We have data and analytics pushed into the business degree. So I think we're going to get more and more generations of young leaders coming in that are already more equipped with an understanding of how to apply artificial intelligence.

0:43:06.2 Todd James: We'll see that become part of the business. So based on partially where I see the puck moving, the Wayne Gretzky, I state to where the puck's going. I think what we're seeing with the education programs and how some of these people are building their careers would indicate that. And two, even what I've seen, when you've got your teams that are closely embedded with the businesses, they open up career paths for those technical leaders into the business, which opens up the possibility to think differently about how you're structured. Now what does that mean? I think data goes back where it was and it's more of a foundational element, sits within IT probably underneath the CIO and they're focused on governance, they're focused on data operations. And it's a different kind of. It's a good role, but it's a different kind of role. Now is that tomorrow? No, but I think transitionally that's where we're headed. I mean, if we're still 10 years out, we don't have companies with heads of marketing, heads of operation, heads of supply chain, heads of financial product that understand intimately how artificial intelligence and analytics can impact their business. I think those companies are going to be in trouble.

0:44:13.8 Satyen Sangani: Yeah, it's really thoughtful. I mean, it's funny 'cause we had another well known CDO, Wade Munsie, who was on, who basically sort of said the same thing, that he thinks these are transitional roles and the foundational elements of data are already going back to the IT world. And he and Ryan den Rooijen were basically like, look, we think the CDO role is fading and the empirical data around that is actually the case in terms of recruitments and the like. And I think that that seems right. 

And what's interesting though is I think what you're saying, which is it's not that that job may be going away, but it's actually really interesting is that the systems thinking that data and analytics roles teach you and sort of the analysis and the ability to piece apart problems, that's really foundational. And this bias towards action is really what you sort of need a layer on. Which kind of brings us back to your beginning point, which is you got to learn, but you got to act. And that seems like a fabulous way of describing your success and why you've been so able to do all these different things that you've been able to do.

0:45:09.4 Satyen Sangani: I mean, so you are now. I mean, this guy that I know, he's been CEO of this company or is the CEO of a company called Zuora, Tien he had this theory about people, which is that everybody, like I look, he's like, I always look at the first job of an individual. And interestingly, in your case, that would have been, I guess, Deloitte, and you would have moved from there in order to go to now back to consulting, where you're going to basically start your own firm. Tell us a little bit about that and what are you focused on and what are you up to? 

0:45:35.4 Todd James: Yeah, I appreciate it. Here's what I'm doing and why, I think beyond kind of the personal interest and the desire to do this, to be able to create an asset and to be able to build an organization.

0:45:46.0 Todd James: I actually think there is a big need out in the marketplace right now, and I think people need help. I think companies are looking for help. And I'm seeing two types of questions come. One, how do I get started? What are my competitors doing? I'm worried about this. I'm going to be left behind. This feels like what I read about at the turn of the last century, when people who didn't rapidly move into advanced manufacturing cease to exist. So how do I get started and what does it mean? The other big question is, hey, I've done this and I can't scale it. I'm not seeing value. I have some use cases, but it's not really changing the business. How can I do it? And I think there are companies that are able to do this, but they tend to be companies right now that are able to. You talked about the CDO, able to pay for that very expensive resource, able to recruit data scientists from leading schools to be able to support it.

0:46:39.7 Todd James: I think there's a need beyond that for all these other companies to be able to accelerate their AI journey. I think it's important for their competitiveness. I think it's important for the future of the US and global workforce. So I think being someone that was kind of in the early phase of how do we start up at a financial services firm? And then how do you walk into a company that has a lot of assets but needs to break through to that scale and increase the value? Having been on both those journeys, I see an opportunity to go out in the market and to be able to have a bigger impact. And like I said, I think it's important for companies to be thinking this way. And so I've got some ideas on how we can approach that and how I can bring some of those resources to companies to help them accelerate their journey. I want to point my career, too. Right. It's a purpose, right? I think the, for me, the energy goes beyond just the economics of the business, energized by the fact that I think there's a fundamental purpose that I can serve by helping people get through these challenges in a way that helps them be successful as a business, but also brings their workforce along, brings their customer base along.

0:47:47.5 Todd James: That's a very important component to this.

0:47:50.7 Satyen Sangani: Yeah, it's great because, I mean, you're naturally somebody who's a teacher, and you're also obviously pretty focused on having real impact. And I think sometimes you have people who are great at sharing their opinions, but may not necessarily see the work follow through. And it's obvious and clear that you've done both. And so it'll be fun to see what impact you have in this new position. Thank you, Todd, for taking the time to speak with us and look forward to seeing you again soon and seeing all the progress you'll have made.

0:48:20.0 Todd James: Likewise, I really enjoyed the discussion today and really appreciated the opportunity to be on the podcast. So it was a pleasure speaking with you. Thank you.

0:48:27.9 Satyen Sangani: What a powerful chat with Todd James. He reminded us that real data leadership isn't just about building smart models. It's about knowing the business, listening to people, and solving real problems. Todd showed how AI, when done right, can change how entire companies operate, improving speed, scale, and customer experience. But the biggest insight, none of it works unless people on the ground understand and trust the tools. He also believes the CDO role is only temporary, as AI becomes part of every function, future leaders won't just use data, they'll own it. And the best data leaders, they'll be the ones helping others lead.

0:49:04.3 Satyen Sangani: I'm Satyen Sangani, CEO of Alation. Thanks for joining me on Data Radicals. Stay curious, stay bold, and we'll see you next time.

0:49:12.1 Speaker 2: This podcast is brought to you by Alation. Your boss may be AI ready, but is your data? Learn how to prepare your data for a range of AI use cases. This whitepaper will show you how to build an AI success strategy and avoid common pitfalls. Visit alation.com/AI-Ready that's alation.com/AI-Ready.