From Back Office to Boardroom: The CDO's AI Opportunity

0:00:03.4 Satyen Sangani: Welcome back to Data Radicals. When I first launched this podcast, I was convinced that a thriving data culture was the key to unlocking business value. But that landscape has changed. Macroeconomic pressures, along with the explosive growth of AI, have shaken up the world of data, casting doubt on both the effectiveness of traditional data strategies and even the role of the Chief Data Officer itself. We've entered an era of tougher questions, tighter budgets and higher stakes. Yet amidst these challenges, there's a unique opportunity to rethink how we leverage data and AI for impact. Today, I'm joined by two exceptional leaders who have witnessed this evolution up close. Ryan den Rooijen, an advisor who has steered organizations through complex data transformations, and Wade Munsie, a seasoned data executive with deep experience across industries. Together, we'll explore what's changed, why it's changed, and how forward thinking data leaders can navigate uncertainty to drive genuine value. Because even in turbulent times, innovation and opportunity are never far behind.

0:01:07.7 Producer: 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.

0:01:30.3 Satyen Sangani: Today on Data Radicals, I'm excited to welcome Ryan den Rooijen. For the past decade, Ryan has advised over 100 organizations on how to effectively harness data analytics and AI. He served as the Chief Strategy Officer for Google's largest cloud consultancy in EMEA, Chief Ecommerce Officer of the Middle East's largest luxury retailer, and was the Director of Data at Dyson. Joining him is Wade Munsie. Wade is a data leader with extensive experience in implementing data driven change on a corporate scale. He brings extensive experience from pharma and logistics and served as CTO at Controlant, Senior Advisor and partner in McKinsey, global, chief data Officer at GSK, and Chief Data Officer at the Royal Mail. He's also a two time global Data Analytics Leader of the Year. Ryan and Wade, welcome to the show.

0:02:21.0 Wade Munsie: Thank you.

0:02:21.4 Ryan den Rooijen: Thank you for having us.

Why is the chief data officer in trouble?

0:02:23.1 Satyen Sangani: So I want to jump right into the topic that really motivated me to have you on this podcast in the first place, which was the article that you wrote on the role of the CDO. You wrote it on your blog. Ryan, can you just take me through the genesis of that article and why you wrote it and what motivated it?

0:02:38.3 Ryan den Rooijen: Yeah, so I think like most great human creative endeavors, it started over a drink with a good friend. You know, Wade and I were sitting backstage at a major data and AI event in London last year, and we were reflecting on the disconnect between what we were hearing on the stage and what we were hearing backstage. On the stage, people spoke about all the incredible successes they were having, how they were delivering all this transformative impact and particularly how, of course, now with the rise of generative AI, that was the topic they were driving within the organization. When you kind of spoke to people afterwards they admitted that actually in reality they were facing a much tougher time delivering against those elements. And obviously the natural question then is, why are we seeing those things? What's happening?

0:03:25.4 Satyen Sangani: And I guess in any endeavor, Wade, you would end up in a situation where there are going to be some successes and some failures. What were the empirical signals that led you to think, wait a second, there's actually really something wrong here. We are seeing a pattern, we're seeing a trend. How do you discern that in the moment?

0:03:43.6 Wade Munsie: The signal was actually really simple, in the end, it was, as we sat down and had lunch, we caught up with colleagues old and new and people we'd never met before. There was a theme that very few people were in the same job they were in the year before. And a lot of them, even though their badges had the company on it, they said, oh, I'm moving on from there, or I'm no longer there anymore, or something's happened. And it wasn't just one or two. It was probably about half the audience. And this was in a VIP lounge for, actually the executive lounge for CDOs, but only probably half the people were in the job they'd been in a few months earlier. And that was pretty telling. We were both quite shocked about it at the time.

And that's where the conversation started. Talking to other people who were there and wondering why. And we were looking down over the balcony at all of the stands down there and all of these people walking around in this AI and data bliss thinking this is a wonderful environment, but actually the leaders who are behind it are going, this is not so great, there's something wrong.

0:04:44.0 Satyen Sangani: And when you spoke to these people, because the outside in signal would, if you were an explainer or an apologist for the data market, you would say, well, it's growing so fast that these people are just getting locked up by these voraciously data oriented employers who just see the value of data and want to get the best and the brightest. That wasn't, I presume, what people were saying; what were they actually saying?

0:05:08.7 Ryan den Rooijen: Wade and I have been, I think, doing this long enough that our network is relatively broad. And so it's not just the data leaders that we talk to, we talk to a lot of recruiters, right? Both the big firms as well as more kind of niche recruiters. We speak to other colleagues, past and present, who hold roles outside of the data and AI space.

And the feedback we got from all those kind of quarters was actually we see a lot of talent on the market from a broader data and analytics perspective. We see a lot of discussions about budgets being pulled or reallocated. And certainly when it comes to the ownership and the accountability for these large transformational programs internally it's no longer the same people from a few years ago. And so it really was this kind of like broad spectrum change.

And I think, Satyen, you touched on something really interesting which is if you're looking at this purely from a market lens, you do think things are very rosy and there's plenty of hiring going on, but most of those jobs are staff engineers for the GenAI majors.

0:06:10.0 Ryan den Rooijen: That's just the reality. And as much as people like to throw data and AI together, there is a massive world of difference between someone who knows how to fine tune very large scale, complex generative models and someone who knows to write SQL to move data from an Azure data warehouse into Snowflake.

Why are data leaders facing short tenures? The challenge of value

0:06:26.9 Satyen Sangani: Yeah, absolutely. So there's been so much promise in data and you know, we all talk about the success stories to the point that you made of this conference, but the failures probably have a lot more learning for data leaders than the successes do. What does the unhappy circumstance look like? What are the patterns that we're looking for? And why do people get let go of so quickly after being first hired?

0:06:49.6 Wade Munsie: There's probably a couple of things. The biggest thing is the reality didn't meet the expectations. So people think data is so easy to manipulate and turn around and the value is just going to appear and that just doesn't happen. You know, if you want to go through a big scale transformation, it takes time.

But I think also traditionally CDOs were in place to wrangle and collate the data and curate the data to a point that it was perfect and that was the ideal for a long time. I think that's probably an impossible task these days with all the different types of unstructured data around there. But also, is it needed?

So if we keep pushing for that nth degree you are never going to achieve. And if you keep pushing for that from a quality point of view and a curation point of view, you forget about why you were there in the first place, which is value. And if you don't get to that value point quick enough, it's very hard to explain why you owed that budget in the first place. Where will that cost when – and it can't be, I think Ryan made a point in one of the articles about all I got was prettier dashboards…

0:08:00.3 Wade Munsie: That can't be the output of it. There's got to be a lot more to it and it should be long term value and partnership with the business. And it's not just about my world as a data officer being about data governance and some dashboard in this area. Much more to it than that.

0:08:14.3 Ryan den Rooijen: Yeah. And I think on that point and then Satyen, it'd be interesting to get your take on this from a vendor's perspective, but there's clearly been a sea change in many of these organizations, right? And maybe you can compare kind of the data scene from maybe five years ago to someone who has an awful lot of money and owns 10 cars.

If someone owns 10 cars, they have a car collection. You don't say, oh, this Mercedes can I use it to pick up the kids or do my grocery run? Because this person clearly has 10 cars because they're just collecting cars. If all you have is one car, then clearly someone in the household is going to ask, okay, but this one car, can I use it to pick up the kids and go for grocery runs? And it's a bit similar from a data perspective where five years ago you talked to some organizations and they say, oh, we've got a massive Hadoop platform, right? Or we're going really big on BigQuery or Snowflake or whatever the flavor of the month is. And nobody even asked, well, what are you doing with it? Because everybody wanted an MDM program, right? Everybody wanted a data catalog.

0:09:15.5 Ryan den Rooijen: And that was fine, of course, until the macro environment got a bit tougher and people said, okay, but now seriously though, what are we doing with all these investments? Why do we have 10 cars? And it turns out that for most households, it turns out you don't need 10 cars, but what you definitely need is somebody who knows how to drive. And a lot of these organizations ended up with 10 cars and no drivers. And of course the CFO's responses as well is, sack the lot of them. The CIO's response is, that's fantastic. I will take those people and fold them into my organization. And I think those two scenarios are definitely ones we've seen play out on a global scale.

0:09:51.4 Satyen Sangani: Yeah, I presume in this analogy or in this metaphor, I'm the seller of cars, are car manufacturers...

0:09:57.3 Ryan den Rooijen: I'm not pointing. You're in the business. You're in the business. I'm not...

0:10:02.5 Satyen Sangani: That resembles reality. I mean, look, I think that even this podcast, which is called Data Radicals, was built on this premise of data culture being good in and of itself. And there was this premise that sort of if we all invest in data and we invest in the tools for data and the people around data, then of course, amazing things are going to happen and we're going to make a lot of money and we're going to be better than our competitors, and we're going to just make all sorts of daisies and flowers appear out of the air. And I think that narrative probably lasted a good 10 years. I mean, really, from the advent of Hadoop to about 2022-ish, I think maybe 2021. I mean that was a belief, and I think it was a belief born out of zero interest rates and lots of new technical innovation. And by the way, I think true of data and true of a lot of SaaS across the board.

0:10:57.0 Ryan den Rooijen: Absolutely. Yeah. It's a field of dreams strategy.

0:11:00.2 Satyen Sangani: Yeah, absolutely. And then everybody's like, wait a second. Actually, I get measured on the investments around this stuff, and the people that I hire and the tools I buy actually have to have some fundamental ROI. And and I think to the point that you made, Wade, the thing is that everybody has this, like, I want to get a business outcome. And then data leaders come in and say, well you've got to do master data, and then you've got to do some metadata, and then you've got to go catalog some data, and then we've got to do some governance, and then we've got to, like, do some data quality and then you'll get an answer. But of course, by the time the then happens, you've died somewhere in the middle of the video game of like, it stops because you basically died somewhere in the middle of data quality because you never were able to get to the outcome. So, yeah, we see that. And I think it's amazing. I mean, I think it's amazing because it's amazing that all of us sort of believed in the myth simultaneously, I’m included.

0:11:53.3 Satyen Sangani: Also amazing, because it's good that we're all getting back to the ground truth as to why this stuff matters and what we're supposed to do with it.

The CDO as a transformational role

0:12:00.5 Wade Munsie: There's another driver you mentioned something there which just triggered a thought was around about SaaS platforms. And I think there's also been a change in that same period where the amount of software as a service offerings have changed how investment profiles of CDOs work. So when I was setting up data offices a while ago, I had an OpEx budget of a certain amount, but most of my big programs were done through capital programs.

As we've moved away from those huge on prem solutions to a license based solution, you're pushing OpEx and spend as you go. That means you're then in the target, the crosshair of the CFO. Because actually that OpEx number is what drives the business. The CapEx can be written off, you kick that can down the road, but the OpEx is front and center. And maybe that's also been a change which has been a bit of a driver. And we didn't actually touch that in any of our pieces, Ryan. But it could be one of those leading indicators actually of where and how we got to where we are.

0:13:08.2 Ryan den Rooijen: Yeah, I think there's an interesting piece here where... And we see this as well in like a lot of these surveys. Our friend Randy just published some of his latest findings. And what's interesting is that there's always a majority of CDOs who say, oh this role is definitely going to be there forever. And we really see it as a board position. Not to be too much of a Debbie Downer, but quite frankly, I have never met a board member who has told me that they see the CDO as a long term ExCo position. Even when I was a CDO and sat in the ExCo, it was very much like we see this as a transformation role. Your role is to kind of move the business from A to B. You're not necessarily here for the long term, which was okay because at least it makes you mission driven. Because of course the fact of the matter is that as these programs and these data capabilities become more important to an organization. Right. I mean, to Wade's point, as the OpEx becomes material as well, different business functions become more interested and tend to want more ownership as opposed to the opposite.

0:14:00.7 Ryan den Rooijen: Right. So for example if you are a CDO building up a data office, nobody's ever thought about supply chain data or customer data. You can get away with owning that data because nobody else cares. If all of a sudden you have a beautiful single customer view with millions of very rich customer records. Like, you bet that the CMO or the CCO or the COO is going to say, great, that's now mine, because it's going to drive the majority of my business. 

And so in a way, this idea of, oh, but we are going to be there for the long term as CDOs, because data is going to become really important, kind of ignores the fact that when things become important, other people also tend to want to own it. Which I think makes a huge amount of sense because you see this with Chief Digital officers as well, right? There's no CMO who says, oh, I don't do any of that digital stuff. That's the Chief Digital Officer's job. It's like, no, like if you are a CMO, right, or a COO and in a modern retail business, you own those capabilities now because that's just how business works.

The 3 types of data leaders

0:14:55.5 Satyen Sangani: Yeah. And I think there was this sort of interesting type. I mean, you guys have this interesting typology where you sort of said there's kind of three different sorts of people who come into these jobs and on some level they bring with them their fundamental biases. Can you walk the audience through sort of what these typologies are and who these people are and how they come to the job?

0:15:15.9 Ryan den Rooijen: I think there's probably kind of three maybe trajectories, if you will. And I think this probably goes for the kind of coming in as well as the kind of going out piece which is maybe even more interesting than kind of where the people come from. Right?

I think that traditionally you'll have these types of governance, right? And the kind of command and control kind of folks, right, who want to catalog everything. They want to have MDM for everything, they want to govern everything. And they're very much kind of coming at it like, if you organize, the world will have created success, right?

Now the reality is that when you had the first data leadership roles emerging probably in the early ‘00s, right? So with the major banks, that mission really was the correct one, right? Because in their case, they needed to meet regulatory requirements, they needed to organize their customer data, etcetera, to meet regulatory reporting standards. But of course, as the data leadership role kind of spread, you kind of saw the kind of like needs, different types of needs, right? So I think that kind of then led to these kind of technology roles, right? These technology CTOs who kind of came in very much with engineering hat to say, hey, data is going to be an integral part of our capabilities as an organization in the future.

0:16:22.2 Ryan den Rooijen: How do we bring in this kind of expertise? I mean, to Wade's point, how do we set up these massive Hadoop clusters, etcetera, and really build the capabilities that we as an organization can leverage? And then I think maybe the kind of third type. And we see this particularly on the kind of the D2C side, right, driven by this whole conversation on customer data platforms and the broader MarTech experience is where you have more of these kind of commercial data leaders come in who have like a very kind of practical lens on, well, I just want to be able to do X with the data. I want to be able to segment this. I want to be able to create a journey a certain way. And what kind of data capabilities do I have? Now, I think the challenge all of these different roles have faced is that generally speaking, a lot of the focus as you mentioned, Satyen, has then been drawn to the technology.

0:17:10.8 Ryan den Rooijen: And it might have been a slightly different flavor, right? The governance technology might be slightly different than the let's build a big single customer review technology, but they still turn into very big programs of work that really focus on the data in itself. Whereas obviously, from a business perspective, the data program isn't the bit that they're most excited by most of the time.

The vendor vacuum, the complexity conundrum

0:17:30.3 Satyen Sangani: Well, and I think the, I mean Wade, you mentioned this expectations question, and I think it kind of gets back to that. But what's interesting about the data role is that the business person hires the data leader. The data leader comes in and the data leader doesn't, interestingly enough, try to normalize or reduce expectations often – some do, but it feels like many don't. They actually are like, no, no, no, no, you're totally right. But to do this job, I need five cars and I need 1500 people and I need this consultant. I need these methodologies, and then I'll be able to give you an answer. And often what's interesting is at least I find that they speak a language that is totally alien and foreign.

0:18:15.6 Satyen Sangani: And in some ways, some of them build organizations that often are there to multiply the complexity. And the vendors encourage this. Like, we all sit there and we're like, of course you have to do, I had this one competitor, I won't name their names, but they were like, well, to do proper data management, first you have to go build a data glossary, and then you have to go build workflows, and then you have to go build some MDM and then you have to go catalog your data and then you have to go do this and you have to go do this and it's like this is a roadmap for me to sell you more stuff that will get you nowhere.

0:18:42.0 Satyen Sangani: But it's interesting because normally what people do is they say to get to success, I actually what I need to do is deliver on this small thing. But that's not, it's seemingly the bias of how people come to the table with this job. And I guess I do wonder why that's the case. Because these are other people who have otherwise been successful, run projects, often delivered projects. Why do you think that is? Like what's the pull is it is the vendors, is it the consultants? Where does it come from?

0:19:08.2 Wade Munsie: I think it's a combination of all of that. And if you, I think Ryan and I have been to a couple of conferences together where we talk about this. People talking in a vacuum like they're on cloud nine. They're all bigging each other up about how great the data industry is in a silo. But if you asked anybody outside of the data industry, they go, I don't give a monkeys about what you do if you don't deliver it, I'll be on your back if you do. I won't give you a high five either. I think we forget our place in the world in an organization, in a company, a little bit too much. And it's because that vacuum we're living in is very, very good, supported by vendors and consultants and whoever else. To talk about this utopia of a data driven organization and what having amazing data can do and leveraging AI and doing all these wonderful things and data literacy programs and yada yada yada. And in the end, unless you're delivering value, the business just don't care. So you can go and accept. I've got a pile of awards behind me over there from over the years, but I've reflected on a lot of them since going great, makes you feel good at the time.

0:20:21.0 Wade Munsie: But who really cared in the organization beyond my team and beyond my peers? It's hard to reflect back on it with and not feel a little bit down on yourself. But I was that person. I created big data teams of more than 200 people, 300 people with BI teams and analysts and master data teams and governance structures and engineers, all there to do the one thing. But it's not how I look at the world now as I go into new roles and I'm thinking about structures of organizations, I want them to be as small as possible. I want to push as much work into the business functions possible. I want to orchestrate and help drive the strategy to achieve value. But it doesn't mean we need to own anything. And I think that's a shift that a lot of us are seeing who are still in the industry. And the people that go in and try and build these huge monolith data teams, I think are the ones who are being left behind.

0:21:27.5 Ryan den Rooijen: You know, I think like many life lessons, right, it's easy to have perspective once you've done it, right? And I think picking for a role as kind of novel as ours, I mean, same thing, right? I mean 10 years or so ago I was like I want to be a CDO, right? So I was like, if I'm a CDO, then all these problems that currently exist are going to go away, right? Oh, if you're an ExCo, then a whole bunch of problems you've got are going to go away. And turns out these problems didn't go away. Turns out that actually titles are a fantastic example, right? Far more important than a title is the question of what kind of mandate do you have in reality? What does your title say about you when it comes to the business, right? If your job is director of making you money, you're probably going to want to meet with me because my job quite literally is making you money. If my job is director of advanced data products and sophisticated AI magic, you're like, this guy's probably overpaid. I don't want to buy him a coffee.

0:22:29.4 Ryan den Rooijen: But it's one of those things where you don't really realize until you've just basically been probably knocked down a couple of times. And I think Wade and I have enough years in the saddle that we've, I think, realized what's worked and what hasn't. And maybe that's also why we took the time to write all of this and get some of these thoughts out there, because it really is frustrating maybe to see all these amazing talents, all these investments, these organizations, and then hear people just be so down on them. And while I think it's absolutely right that questions are now being asked by CFOs, by CEOs, by wider leadership teams about the role that data and AI should play in the organization, and questions are being asked about ROI and about value and conversations arguably should have happened years ago, but find are happening now. That's not the same as saying, I believe that everything out there is crap. There's an awful lot of people that Wade and I know who have lost their jobs in the past 12 months who are fantastic leaders and fantastic people. And that really is heartbreaking to see because, as you said, Satyen, it so often comes down to a lack of communication.

0:23:36.0 Ryan den Rooijen: You know, one of the great challenges, is the illusion of communication has taken place. I think it was a... I'm probably misquoting Bertrand Russell here, but it's exactly that, right? People feel that, oh, we've got a data strategy, so the business understands me. And then people lose their jobs and they go, but I had a data strategy. It's like, well, yeah.

Redefining data culture as value driven 

0:23:52.7 Satyen Sangani: Some execution would be nice with that. So if this podcast is, like, about data culture. It's not that data culture isn't valuable. There's this myth, though, that the culture is built based upon process and behaviors and inputs. And actually, the culture is really just built on winning. So winning is like achieving business outcomes. And if you're a data leader and if you want to win, go forget about which tool you use, forget about, like, what process you're using, forget about whether you've implemented data products or not or have the latest tool or whatever. Just, like, go in, go produce some value. You'll probably get some more budget to go produce more value.

And along the way, you might want to go buy some tools, but the script has been flipped in the wrong direction. And I think as somebody who sells one of these tools we're pretty religious about this. We want to work with data leaders who know where they're going. And one of the things that we often have is that companies come to us and they're like well, we want to do some data lineage.

0:24:54.3 Satyen Sangani: For some reason, that's the tell. That one thing is really interesting because then you're like, well, why do you want to do this? Well, because we want to do lineage. I'm like, no, no, wait. We just... I just asked you a question. You answered the question with the same thing that we just told you. But I think there's this desire for perfection because everybody wants to organize it, everybody wants it to be better, but interestingly, it's never going to get better. And what I found is that the organizations where they do the best stuff with data tends to be where the data is the biggest mess.

0:25:24.7 Ryan den Rooijen: I mean, Wade made an excellent comment earlier when he spoke about the vacuum chamber or the vacuum or the echo chamber and a lot of these kind of data conferences. And I think the challenge, sometimes there's this emphasis on data culture, culture can also lead to a cult. And you know, you go to some of these data events even being organized internally where they want to, "promote data culture." And 90% of it is show me your Tableau dashboard and you're like, what are we doing, right? Why do we need to talk about the latest ML research papers? Why do we need to talk about like how you can do like the 20th type of heat map in Tableau? These things are absolutely not relevant to anyone in the ExCo. And again, not to be too reductive about this stuff, but I think there's something such a lack, I mean Wade mentioned this before as well, such a lack of awareness that as a new investment for many of these organizations there really is an onus on people like ourselves to prove ourselves to the organization.

0:26:22.6 Ryan den Rooijen: And so I think the biggest kind of data culture challenge is really how do we make ourselves relevant to the day-to-day of the employee. How do we make sure that if somebody is on an oil rig or in a store or in a call center, on a trading floor or in a lab, they are going to do something different because of us? Because if they're not doing something different because of us, then honestly we don't deserve to be here. And that's an awkward conversation to have because it's much more fun to talk about some of the latest GenAI benchmarks and look at beautiful Tableau dashboards but...

The CDO role as training for other executive positions

0:26:55.2 Satyen Sangani: Yeah, so these data leaders you mentioned, value people are trying to didn't quite achieve it. Data leaders therefore are getting fired. One of the conclusions of your articles was that now the data ownership is moving into, in many cases, the CIO or the CTO. Is that the answer, is the answer that this just needs to be another technology initiative?

0:27:18.8 Wade Munsie: There's data ownership and then there's owning the data part of the business. And I had a lot of reflections on this. So at McKinsey we obviously advise a lot of clients and I worked with people on data culture, data strategy.

But then I listen to the tech guys and the suppliers and the vendors and they're all talk about, well, you need to do a data mesh and you need to have a domain based architecture and you need to do self serve and you need to do this and push everything back to the business. So what's left for the data leader? And if I push everything back into the domains, all the governance goes back, all the self serve goes into the business units. What's left other than just orchestration and setting some policies or standards and maybe trying to glue them together? Certainly if I'm in a company of 100,000 people and my data team was 200 or 300 before and I've pushed them all back into the business, I'm left with what, 20, 30 people maybe who are running better orchestration and that strategy piece and gluing it together for standards purposes. Well, that doesn't deserve a C title for me.

0:28:30.0 Wade Munsie: Try and tell the COO who's got 90,000 of those 100,000 people that you deserve a C-suite and a seat next to them at the table because you've got your 20 or 30 data people. It just doesn't add up.

So where do you go next? Well, either you become one of those people and you start to put yourself in their shoes and think about beyond the data to think about value and think about their business and what drove them and then go backwards and go, ah, now I kind of get it. And this is how we could have helped much better in the past. But I'm going to do that going forward because I want that job. I want to see myself in one of those roles going forward.

Is it realistic? I hope so. I really do. And I've seen the transition in some places, not a lot, but I have friends that have been CDOs who are now COOs, CDOs who have become CMOs and CDOs who have become CIOs, which I did as well as a transition. But you've got to be bold in that invention. You've got to think about how your thinking becomes very T-shaped and you're getting outside of your data echo chamber and really focusing on the business outcomes and why they come to work or why you come to work.

0:29:45.1 Ryan den Rooijen: Absolutely. I think a big reason why we wrote that article that just got published this week in MIT Sloan Journal about future of the CDO role, particularly looking at where CDOs can go from here, is because I think in so many roles that succession conversation is just part and parcel of your development. Right. I mean, if you are a incredibly strong finance business partner, there's probably going to be a discussion about the head of finance role. There's going to be some type of conversation about you. Do you want to be CFO track? Right. Do you want to move into like GM role, maybe divisional management, etcetera, you know, same thing for marketing leaders. Whereas on the data side, it's kind of just been, yeah, your data like. And again, whether your title is like head of data, director of data, or CDO, it doesn't matter because they're all used interchangeably. Perfectly frank. There's plenty of CDOs with zero reports. And a very insightful comment that a headhunter friend made a couple of months ago is that within their firm, the conversation about data leaders was always focused on that data element, right? Who has the latest expertise, who has.

0:30:45.9 Ryan den Rooijen: Has the right type of data tools that they know how to wield. And it never really focused on the word leader. And so I said, well, have you ever had a succession conversation, right, about a CDO? And they're like, no, actually it's not something that we did.

Now, I'm willing to bet that for the vast majority of organizations, certainly all the ones that I've been in, yeah, data people are just data people, right? Like, doomed to kind of stay in that specialist box. And to me, that is, that's just such an unbelievable wasted opportunity. Because to Wade's point, do I believe that journey of the business functions, integrating these data capabilities, taking more ownership, do I think that a mature data function should have engineering capabilities that sit as part of the CIO or CTO's remit? Absolutely. I think that's exactly the way things are going to go, and they should go. But then, to Wade's point, if you are a veteran data leader and you're sitting there on your shrinking little island, do you despair? I think no.

0:31:42.6 Ryan den Rooijen: I think let's get excited about all the opportunities that are next. Because if you look at, for example, marketing 10 years ago, you could be a CMO and just be an expert at brand building. You know, nothing else really mattered. Now, if you're a CMO and you don't understand what a CDP is, right? I doubt you're going to get the job right.

Same thing for COO, right? Back in the day, you could say, well, I'm just really good at negotiating with our suppliers. Now, unless you understand how supply chain analytics work and you can really squeeze every percentage point of efficiency out of it, you're not going to be in that job for long. Not to mention, of course CEOs who are under pressure to be much more data driven, not just in how they set strategy or how they relate to the street, but also in terms of how to think about monetization, product creation, etcetera. And so to be frank, do I feel that data leaders such as ourselves are in a place where we should be stepping into those kind of, I mean, effectively true C-suite roles? No, because I think most of us haven't had that type of investment.

0:32:42.1 Ryan den Rooijen: But I think with just a little bit of love and watering and sunlight, I think we could have a fantastic crop of future executives.

Data leaders as systems thinkers

0:32:50.2 Satyen Sangani: Yeah, I mean, for any functional leader, I mean, as a leader of seeing executives, there is this bias towards thinking that your first team is the function that you live in. So a sales leader is a salesperson first and foremost and wants to optimize the sales function. And that generally does pretty well, except when you think about the idea that companies are systems and if the system doesn't do well, then none of the component parts can do well. And so you're always, at least, and at least for me, and I'm always looking for people who are not functional leaders, but who are basically systems leaders.

0:33:31.7 Satyen Sangani: And I think the data role has this one incredible advantage, which is that often data leaders are system thinkers. There are people who are trying to optimize the system, but they've been focused on the wrong system. What they're trying to optimize is the data stack within a company. What they should be optimizing is the system of the business. But people forget that. And it's super easy to forget that because you're day to day is so stuck in the minutiae of did this ETL routine turn off or on, or do I have the appropriate data quality? Or did this report actually work as opposed to like, wait a second, did 90% of the things that I did yesterday actually produce any value? And you know, Wade, that gets to your sort of point of hey, what did I do? Where did my time go? And I think that's a... But I think these skills are really quite useful if you just put them at the right place.

0:34:09.5 Wade Munsie: Totally. And look, I mean, I had this conversation with Ryan only a few hours ago about, about systems thinking and about complex systems require people that can provide the connection between them. And that's the perfect role for a data leader. We talk in a currency which flows between systems, between complex systems.

But I was reflecting back on the team structure and the life of a CDO and to Ryan's point about did we want to be a CDO at one point? And when I first became a CDO, that was an amazing day. I loved it. And then I became a CDO at another company even bigger. And it was after being CDO at a couple of Fortune companies, you go, now what? You know, BAU do I want to be worried about that BAU stuff? So that's when I took the leap into to go to QuantumBlack and McKinsey. Well, does that necessarily tick the box in being satisfied? Some people, it wasn't for me. But then CIO role came up and that was even different again. But now I've stepped back to, into data and in a new role. Now I'm back to leading a data transformation.

0:35:19.3 Wade Munsie: But I've been very honest with them saying, this doesn't mean you need a CDO. If I do my job right, you probably won't need a CDO at the end of this. And that's okay. And I think we've got to be really honest with ourselves about where we're going and going into a job, because I'm sure people will be listening to this and go, well, Wade's written all of that stuff out there about the future of a CDO, but now he's gone back into a CDO role. It's like, well, no, I haven't, actually. I've gone into a data transformation role, which, if I do it right, will mean there is no need for a CDO.

And being honest with yourself about that future is the first step to working out, where next. And I see myself much more now as a strategist looking at systems, thinking of complex data problems across an organization and providing that intersection between them all to drive value, to drive the company forward. That's really our only purpose. It's like it's the only purpose we were ever there to do in the first place. We just lost our way somewhere along the line in the quagmire of vendors and ecosystems and consultants.

0:36:24.3 Ryan den Rooijen: I think. I mean, I think that's such an important point to emphasize because if you're a vendor, if you're a consultant, you want your client to have a CDO because that's the person that will buy your software that's going to sign off on whatever you need to sign off on. But to Wade's point, taking a step back, what does success look like? What's the North Star? And occasionally people say, oh, you guys are so down on CDOs and data leaders. And we're like, actually, no, right. We're just bullish on those profiles having a much broader impact. But most people in the street don't care about data leaders.

0:36:56.5 Ryan den Rooijen: If you say I'm having a data leader party, right. Most, even most of my friends aren't going to show up and many of them are data leaders. So I think we've kind of shot ourselves in the foot a little bit. Whereas if you go to any executive, right. If you go to any board and say, hey, is leading with data important? Absolutely. Right. I sit on the board of one of the UK's largest nonprofits. And guess what? One of the biggest topics that we discuss in the board is data. Obviously in a very it's not so much of monetization and much more about how we can deliver services more, more effectively. But this conversation is everywhere. It's just how do we open it up and how do we frame it in a way that really gets audiences to engage? And like I said, Satyen you've got this amazing background when you are a data leader where you tend to have to think in terms of systems because you have to work transversely across the organization you're dealing with very kind of complex stakeholder usually.

0:37:52.0 Ryan den Rooijen: So in theory, the ingredients are there for you to have that much broader scope of impact. But it's something that people need to consciously acknowledge. To Wade's point, there needs to be maybe a bit of an awakening and people have to reflect on, do I see myself as a CDO or do I see myself as someone who's going to deliver that transformative impact using data? And those two things are absolutely not the same thing.

0:38:15.0 Satyen Sangani: Yeah, by the way, I share your optimism and I think the observation that the current state doesn't work is real and true because like, nobody wakes up in the morning wanting to like. Data isn't a thing that needs to be led. Data is a means to an end. And yes, like a CEO, if they're trying to build a data driven organization or lead with data, as you put it, is trying to basically build a culture of trust and transparency, which are great things and data is a means to that end.

But I do think that so you can lead with impact and that would be the path for somebody who get, who wants to get to the C-suite. But if you're a great data person who really just likes building things, maybe you should go to the technology side or if you're a great person who really loves organizing things, then maybe you should go to the operation side. Or maybe if you're somebody who really loves to do analysis, then you should become an ML leader who you know, seem to get paid pretty well nowadays. So there are lots of paths, and the fact that the CDO path may not be the one to focus on actually might be positive, not negative because you're that was like a Pyrrhic victory.

Should data live under IT? The maturity question

0:39:21.2 Satyen Sangani: And what you really want to do is actually have more impact and figure out your path and way to forward.

So I'm curious if you think data should live under IT or if that's where it's going. Maybe I'm not even sure that the CIO in some cases knows even what they're inheriting. You know, the CIO has a technology portfolio like they have a portfolio of sort of applications, they have a portfolio of infrastructure, they have to do security. Security in particular is like a very, like it's insurance policy thinking. And then data is this annoying odd thing. What is your perspective on it or what are the pitfalls if they do chase to take on that ownership?

0:39:56.6 Ryan den Rooijen: Wade and I've talked about this a lot, so he can talk about the broader philosophy. But I say I think the key element here to start by distinguishing between data as a whole and key elements of data. Right. Because the reality is if you have a very large scale enterprise data platform, should that be owned alongside all the other enterprise platforms by the team under the CIO or CTO? Of course. Right. If you have a cybersecurity function, safeguarding the broader estate and data products are yet just another application that sits alongside other applications, should it be owned? Absolutely. Right. Is there any CDO that should be running a service desk and taking tickets if you already have end use support? Of course not. But I think that's a very different discussion than I think the one that we're currently seeing. And again, Wade can talk about this where some organizations people are going, ugh, this wasn't worth it, let's just get rid of CDO and stick the rest under the CIO.

0:40:49.6 Wade Munsie: Yeah, that becomes down to maturity. Right. So if as I mentioned before, you become in some sort of parallel universe, amazing at being a domain based architecture and you pushed all your data domains back into the business and the business have the right stewards are allocated to them and they're managing their data really well. Wonderful. Self serve is running great. You've got your data scientists embedded in the business with an MLOps platform running everything, everything's coming along. What am I handing into IT? Well really I'm just handing in, run the platform for me, feed and water IT, and tell me when there is a major path that needs changing or I need the expertise to decide the actual data strategy beyond that. Beyond that, I don't really care what you do with it.

0:41:47.2 Wade Munsie: Look after it, make sure the plumbing works and make sure someone doesn't switch it off when they leave the office at night. That's what you really want IT to do. IT do IT really well. Data doesn't need to, as Ryan said, you don't need to replicate it. The challenge though with maturity is if you are not mature enough then creating data products is very, very hard and you need data professionals to help curate the product story because the business aren't going to be mature enough to articulate those problems down and then you end up with a world where you've got half baked products that never make it to production and never see the value.

0:42:23.1 Wade Munsie: Which is when you get to the point of where we are with a lot of CDOs today is they've got their own stakeholder groups working with the business. It's very confusing because IT also have business partners and then there's confusion there. Then you're fighting with IT all the time because actually you want to put stuff into production. They go, well no, we own that platform or security won't let you do it. You have not gone through the same due diligence or you've not done this or you've not done that. It's not part of our enterprise architecture plan.

So there are puts and takes there with maturity becomes responsibility. But then you've also got to relinquish and be able to happily give it up. Say, you know what, I've done my bit of the transformation. Please take this and run it for me. That's your job. However, when it comes to the strategy side of data, please let me do it for you, or at least with you, because that's not your go to either. You know, IT don't talk in terms of the values facing most companies. They talk in terms of cost and they're there to reduce the overall cost of technology, to keep a service running, to be safe and secure.

0:43:29.8 Wade Munsie: It's a different purpose and it's a different outcome when you've got different purposes. And as long as you understand that you'll do well. You can't just dump it over the fence to IT because you wanted to fire the CDO and give all of those analysts and scientists to a technology organization because they'll just leave.

0:43:45.9 Ryan den Rooijen: Yeah, and I think there's an interesting point here as well, which is, it's very important to say that when we talk about IT, we're talking about IT in really the kind of like the most kind of classical business school sense of what IT is, because in many organizations now, modern organizations, you'll have, for example, a CTO. Now, a CTO might own that IT part, but they might also own, for example, product management, which is very much about customer value, etcetera.

And so another perennial topic for the fringes of these data conferences. Oh, well who do you report to? You know, and if you will say I report to CIO or CTO, you get groans and I'm always like, whoa, whoa, whoa, forget about the title, right? Let's talk about what functions actually exist. Yes. If you as a data leader have been shoved under a CIO who purely is all about IT and procurement and offshoring and whatever, then maybe there's a challenge there, as Wade pointed out, right. Because how are you going to have to space, have those value conversations with the business? What's your scope to set organizational strategy? So fair enough.

0:44:43.7 Ryan den Rooijen: But like I said, there's also plenty of cases where you've got the most amazing, these amazing unicorns. Right. A lot of innovation and value creation sits under a CTO, right. Or CIO. And so it's, I think it's very important to not be too reductive when it comes to titles and really think about, as you say Satyen, that the kind of system level of how does the organization as an organism function and where those elements of like value creation versus delivery, etcetera, strategy all sit.

How will AI impact data teams?

0:45:09.9 Satyen Sangani: Yeah. And also if you're a CIO, I mean, if you want to be successful, then the last thing you want to do is inherit the bag of expectations that the CDO failed to deliver and then be like, oh, great, I've got this entire team under me and now I'm the new punching bag for the thing that the old punching bag got fired for. That also seems like a recipe for disaster. And so on some level, then that CTO or CIO has to be pretty authentic about, like, what do they truly want to do? Do they want to be the person who owns the technical architecture and the infrastructure for data, or do they want to have a more expansive relation into the business? And if they want the latter thing, that could be a huge win to get to the next level. Because if you really have that ambition, but you just have to realize that's what you're delivering on. We, you know this wouldn't be a podcast in 2024, like we wouldn't even be able to allow to print it if I didn't say the word like AI and agents and so I've said the words.

0:46:05.4 Satyen Sangani: You both watch the industry, see a lot going on. What's the impact of AI and agents on the world of data and what's actually happening? What do you expect to have happen? And are you guys optimist about this or skeptical or where do you stand?

0:46:19.6 Wade Munsie: I'm very optimistic about it. I'm also skeptical about it. I think I'm a bit of a realist and a skeptic, but there is a huge positive here. Right. The first thing is back at that conference we were talking about and I was standing on the balcony and looking at probably 100 stands of companies with “dot-AI” written after them. If I go back to that same conference this year, I don't know how many of them will still be of existence.

So you've got to take a lot of it with a grain of salt. I mean if every one of them worked really well, I can't imagine the ecosystem you would have in your company. In saying that there is a lot of good there too and, and whether it be about how and actually Ryan and I talked about agents before as well about being evolutions of microservices and where are we? You talk about these different tools and some of them are just rebranded things of stuff we had before. But there is a lot of good there. The copilots of the world are fantastic. I've got one on my other screen here doing transcript as well as we're talking because I wanted to be able to see how things panned out and run it through a gen AI model later to take the highlights from it just for myself.

0:47:29.8 Wade Munsie: Right. That's useful. It's useful in your day to day if you're a bit of a closet geek like myself. But in a day job it is useful. The amount of performance or optimization people have got by using copilot so you use your tools is great. Is it dangerous? Probably a little bit. People become a bit lazy sometimes. Lazy code is a real thing and I had to have a lot of governance around copilots in my previous CIO role. But there is an upside to it for sure. I see value pockets everywhere but I'm skeptical that we're trying to run a little too fast sometimes and people are being sold the dream. But there's not a lot of AI running or especially GenAI running at scale in most organizations. In fact, very few from what I can tell.

0:48:21.8 Ryan den Rooijen: Yeah, I mean, not to be that guy, right, but I mean, I think back in 2018, I think '17, '18, I did a keynote and we were in the middle of the whole deep learning hype at that time. And you know, I got all these questions afterwards and I'm like, okay guys, like, this is great. I'm like, but how many of you like have even used like old school ML, right? Like SVMs or something, right? Like basing these kind of classifiers, right? KNNs, whatever. I mean, and you know, a lot of people in the audience are like, oh no, we just want to do, we just want to do the deep learning stuff. And I'm like, yeah, I know, that's why I put it on the slide and stuff. I'm like, but I really want to get into like how do you actually use AI in a much more practical sense? And then I feel like if you go to a lot of these organizations that are now talking about wanting to go agentic, you say, amazing, how are you doing segmentation? How are you doing optimization? A lot of these folks haven't even started with the most basic tools in the toolkit, right? And it's kind of insane if you look at the amount of compute that's now required for some of these models.

0:49:22.9 Ryan den Rooijen: I mean the fact that people are basically using these incredibly sophisticated and resource hungry tools to solve problems that could effectively be solved with 3 lines of sklearn and it could have been solved about 10 years ago. And so I think it's important to maybe distinguish between almost three elements, right? It's like, yeah, there's AI stuff and there's a huge world of AI as we all know, and AI tools and algorithms that people should be applying across their organizations. And I have never walked into an organization yet in my life where they're even using 50% of the stuff they could be using, right? So, and I think also boards stress about AI. It's like, let's not begin by stressing about agentic, right?

0:50:08.6 Ryan den Rooijen: Let's actually figure out how we get people to stop classifying manually or using a spreadsheet or like a Visual Basic rule and start using some SVMs, please. So that's one. Step two is then, okay, GenAI has obviously introduced a bunch of new capabilities that didn't really exist before, right? I mean, yes, you could do summarization before, but it was a huge pain and it took ages. Now, summarization, you can do really, really well.

0:50:24.5 Ryan den Rooijen: Right? You could do multimodal entity extraction, etcetera. Right. So whether music and PDFs and all these kind of things, that's really powerful. But again, I see that as kind of like adding tools, that first set of AI tools we already had. So I think organizations start there.

And like, yeah, then there's now this new conversation around agentic. But again, as Wade said, like with sophisticated integration architecture and microservices and all kind of things, those are very much a kind of like later stage maturity conversation. Like organizations that don't even have their customer data or product data or supply chain data or whatever, or even have, I don't know, standardized date definitions. Right. They shouldn't be thinking about like hyper sophisticated microservices making decisions based on all these different dynamic rule sets because quite frankly, their dates don't even make sense. Right. So like have a little bit of perspective and start with the right priority. So am I excited for agentic? I mean, at the moment not necessarily, because I think there's still a huge amount of value to be had from a lot of the existing kind of capabilities, plus the kind of vanilla GenAI experience.

0:51:29.2 Satyen Sangani: I'm smiling because we're like, literally, I don't know, like 72 hours between putting out a press release from Alation calling ourselves an agentic platform.

0:51:38.4 Satyen Sangani: It'll do wonders for your market cap.

0:51:44.0 Satyen Sangani: You know, it's funny, I have had you could call me a technology luddite. Like I actually, despite the fact that we've been called a machine learning data catalog and like there's been lots of times where people have said to me like, oh, we got to do AI and like we can't even do the manual thing well, like, what the... Like, why are we going to explode that with more AI or. Or you know, like people come to... We had this feature last year where people. We launched a capability where it bulk-described using an LLM and you could get some fidelity out of that. Like you could actually get. But you were taking structured metadata and turning into unstructured text, ironically enough. And that seems like a weird thing to go do. Although there was utility to it to a bit. And so I've not... And then people want to do that at scale where they like bulk multiply it. I'm like, we're just producing garbage at scale. Like you actually have to look at this stuff and like revise it to make it work. But I will say the one thing that I think about these agents that's interesting is that it, I think it's going to change the interface for how we deal with all of the software.

0:52:41.9 Satyen Sangani: And the thing that I'm personally excited about is that to do this data management stuff or to do anything, you have to know things like Alation, you have to know things like data governance best practices, you have to know the policy, you have to know the context in your organization. And if you can digest all of that knowledge and allow the agent to prompt you through the actions, then you can create something that allows you to be met where you happen to be. And so to me the thing that's really exciting about these things is that they can bring a lot of knowledge to bear at the point that the user is making the decision. And so it could be the case that I think there's a, for somebody who's tried to solve this with the old tools, the standard SaaS tools, I feel like this gives us another tool that's pretty exciting.

0:53:25.7 Ryan den Rooijen: So I think you raise an interesting point right on the interface side, like, absolutely. I think there's a lot of potential there. I think the key thing is that there's a lot of capabilities that, that on that front aren't necessarily hugely novel, right? So when I was working at Google over a decade ago, like we had an internal tool where I could talk to my phone, it would in the background pull the relevant tables, data definitions, whatever, run the calculations and give me the results. So I could literally I could say, oh break down this ad format, performance, this versus this for these groups of clients across these marks, etcetera and it would, it would parse it. Now of course there was a lot more work that had to go into building the kind of ML that sat behind that, that tool of course than it takes nowadays. Right now anyone can use a tool like Cursor or custom GPT to build their own models. But even if you look at things like chatbots, right, like contact center, AI and these kind of capabilities have existed for years, how many businesses have a good chatbot, right? So what I think is going to happen is in the SaaS space, I think it's a different story because you can basically build a really tight, well defined product for a particular niche or for a particular audience.

0:54:37.0 Ryan den Rooijen: But again, and it's just like Apple, Apple can engineer world class hardware because that's the thing they do. And like I said, there'll be SaaS businesses that will engineer world-class agentic experiences because that is what they either do or are going to do.

But I think for the average organization who can't even get a chatbot working using technology that's been around for last 10 years that doesn't suck. Right. I'm really skeptical that we are going to see this wave of change as some of these vendors talk about. I mean, I mean returns, processes. Find me a chatbot that can handle returns well. Right. It's still pretty terrible on average. Right?

0:55:16.6 Satyen Sangani: Narrow context is really important.

0:55:18.4 Ryan den Rooijen: Absolutely.

0:55:18.9 Satyen Sangani: Narrow context is very important.

0:55:23.4 Satyen Sangani: Well, this has been everything that I expected it to be. So you guys, even on a Friday night before beers, lived it up to all expectations. This is amazing and thank you for withholding that. Or maybe you've been drinking while you know before, so. But either way, it was a lot of fun guys, thank you for taking the time. I know this will be a really fun episode for everyone listening, so appreciate it and we'll have to have you back on at some point soon.

0:55:42.2 Wade Munsie: Excellent, cheers.

0:55:44.8 Ryan den Rooijen: Thank you.

0:55:48.4 Satyen Sangani: That conversation with Ryan and Wade gave us a clear and honest look at the state of data leadership today. It's obvious that the era of endless optimism, when leaders expected perfect data and instant value, is over. As Wade said, chasing perfection can distract from what really matters. Delivering business results. And as Ryan pointed out, buying lots of data tools doesn't help if you don't have people who know how to use them to create value. But this isn't the end of data leadership. It's a chance to evolve. Data leaders who focus on business goals, learn to work with new technologies like AI and rethink their role can still succeed. It's all about adjusting expectations, staying flexible, and focusing on impact. I'm Satyen Sangani, CEO of Alation. Thanks for listening to Data Radicals. Stay strong, stay curious, and most of all, stay radical. See you next time.

0:56:46.8 Producer: 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 white paper 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.