Connecting Business Strategy with Data to AI Initiatives for Maximum Impact

Speaker A: Hello and welcome to the Business of Data podcast, brought to you by Corinium Global Intelligence. In this podcast we talk to senior executives, thought leaders and experts from a range of industries and departments within large and small organizations across the globe. They share stories and experiences that shape their passion for data and analytics and form the future of our industry.

Speaker B: Welcome to the Business of Data podcast. In today's episode we're diving into what it takes to connect businesses to the benefit of data and AI transformation and achieve AI readiness across the enterprise. I'm excited to have uh, our uh, guests with us today. I think there might be two of the best people to discuss this topic with. Uh, first we have Joseph George, CEO of Dufresne. Dufresne is a prominent UK data consultancy known for its cutting edge data management and analytics services. Welcome to the podcast, Joseph. How are you doing today?

Speaker C: Thank you, Gareth. Uh, doing well, uh, and I'm looking forward to this conversation with you and uh, discussing all things data.

Speaker B: That's great. I was hoping you might like to tell our audience a little bit about yourself and what it is about data and analytics and AI that makes it such an exciting field to work in today.

Speaker C: Yeah, uh, I'm sure like most people stumbled into this by uh, uh, I would say conscious accident. Um, my background's all in been in tech and engineering and consulting, uh, my working career and uh, Dufresne and the data world and AI world felt like a natural extension to my background. Um, I can tell you more about it as we go along.

Speaker B: Absolutely. Well thank you so much. And our second guest is Catherine Masters. She's Director of Pricing and Data at Covey Insurance. Coveya Insurance is the UK subsidiary of Coveya Group, a European leader in insurance and reinsurance. Covey Insurance offers a wide range of home, motor and commercial insurance backed by great service. Katherine brings a business led perspective to uh, discussion today, which I think is so important. It's something we hear about all the time at our conferences in my conversations with data and analytics leaders, how to get at that value and bring data and uh, the business side together. Welcome to the podcast. Katherine, would you tell us uh, a little bit about yourself and why you think it's such an important issue in business today.

Speaker D: Thanks Gareth. Gareth. And it's excellent to be here as well. I love talking about data and analytics and AI, so this is a perfect opportunity and I know we're going to have a good conversation um, as well. My background is um, is very insurance focused. I've been in the insurance industry for nearly 15 years. Done a variety of different, uh, roles, mainly, um, analytically focused around pricing, uh, and data, um, uh, which means that the role I do at the minute, where um, I get to lead those two functions on behalf of Coveyor Insurance, um, uh, is very fulfilling. And, um, I think I've got some good examples to share, uh, with the listeners today about, um, how we're succeeding in what can be quite a complex but also exciting world.

Speaker B: Thank you so much. Yes, there's a lot to get into, so I'll come to you. Catherine. To begin with, when we talk about data and AI readiness, it often involves multiple layers of an organization. So from a business perspective, what are the foundational elements for a company to start a data and AI journey effectively? And how do you turn an idea into a reality?

Speaker D: Yeah, and um, there's no straightforward way to do this. There's lots of ways that you could approach it, but I think about it in three phases. Firstly, um, there needs to be a desire for change. Uh, fortunately, if you're like me and you work with, um, analysts, uh, and people who use data or want to get answers from data on a regular basis, you'll know that they're not shy about being vocal about what they need and what they want. Um, a desire for change and identifying that opportunity, um, usually forms, uh, the first phase M and ideally you might get some good insight from those people as well, what's working, what needs changing, and that can begin, uh, the foundation of your approach. Secondly, there needs to be a case for investment. So if you've defined your vision, you know what it is that you want to achieve. Um, getting the funding for that can, uh, be uh, another hurdle to get over. Uh, and certainly the success that, um, I've seen work well in the past is to structure a program, uh, in a phased delivery where you get quick wins early and begin to kind of feed that case for investment with a more natural momentum. I think gone are the days where executive committees will sign off huge, uh, digital, uh, and data transformations, um, without having a certainty of return on investment. And having a phased delivery can certainly help with that. Um, and then lastly, I think empowering your people can really help you turn that idea into reality. Um, and, uh, one of the ways that we've approached that at COVEA is to lead our data and AI transformation out of the business. Uh, and that means you've got some really fantastic people who know what they're doing, um, both from the business and technology aligned around a vision who can, with support, lead that change which ultimately can improve adoption and the value that you're able to generate from there.

Speaker B: Yeah, I think that's fascinating. Um, so when you're looking at that kind of phase delivery, what are the benefits of providing those quick wins early in the process? From a business point of view, it's

Speaker D: just the tangible nature of them. So, um, you can actually have a really big impact in a very short amount of time. Um, and there's different accelerators to utilize to do that. And it becomes a, um, much easier journey when you're showing that continuous delivery. And I think that's also the kind of expectation now when, when companies are going through changes like this. We're not patient enough to wait until the big pot of gold at the end. We need to be, um, showing delivery and value, um, throughout that journey.

Speaker B: That's great. Thank you. And Joseph, what do you think, you know, thinking about these foundational elements, you know, the beginning of the journey, from your experience working with clients, uh, and in the, in the data industry, what do you think are some of the most foundational elements there?

Speaker C: Yeah, I believe, and we as business Dufresne, uh, believe that everybody can do a better, uh, can do better in their job no matter what it is if they had access to the right information at the right time in an efficient manner, no matter what function or operation you sit in and what industry you sit in. But yet most struggle with that. I think that's the core essence or the lowest common denominator of what I think a, a foundation needs to achieve. Um, so if you look at what you do with data, there is the what happened, using data to see what happened. There's the data to understand what is happening, what data we've got, and then to look at what will happen, which is your more advanced AI piece. And most organizations, uh, uh, from an enterprise level or an organizational level, in terms of maturity, is still grappling with the what, what has happened in the past, the historic bits, and what is happening right now with the data we've got and trying to make sure that they can get access to that information in the most efficient way. And it's the right information. There are pockets of experiments of what will happen in the future and there are definitely some successful use cases across various industries, uh, for that. Uh, but I think as a whole the industry and the world is still grappling with the foundational elements of the basics. So that's how I would see that mature take off of data.

Speaker B: Yeah, absolutely. And you mentioned having access to your data there. And I really think that data governance, data quality really play crucial roles in data and AI transformations. Um, at defrain. How do you approach these aspects, how do you advise your clients about them and uh, do you have any best practices that you can share with our uh, listeners in regard to those?

Speaker D: Yeah.

Speaker C: So as Katherine's mentioned, tangible benefits and phase programs and showing business value quickly. Um, this is one of the very difficult areas to show a business value when you talk about data governance because it's not something that lets uh, everyone get out of bed in the morning and excited to talk about data governance. And it's very difficult to show a tangible business value. So I think most organizations and the various sectors go through cycles of the prominence of this. There's a link to regulation. Um, regulation demands it in many cases like in financial services. Um, but in other cases something has to go wrong which could be an internal audit related activity or an external auditor pointing out to someone where these things typically have um, in the past taken uh, a lot of notice. Now that's all completely changed and disrupted in the last 12, 18 months with AI engine AI and everything going on and the art of the possibility with that because everyone's now realized that without proper controls, proper governance and good data, uh, you can't make the most of AI. So therefore uh, these areas which are typically overlooked, um, or used to be overlooked in most organizations are getting a tremendous prominence and investment. In terms of approach, I wouldn't say there is a one size fits all approach. Everyone looks at it in a different way. Uh, they look at what benefits they try to get out of it with different horizon lenses. But if I was to summarize some of the best practice we've seen and how we work with many organizations is there is a top down and a bottom up approach to this. Uh, when I talk about top down I said do you have a policy? Do you have, what are your business objectives, what's the business strategies, the business and the leadership and the exec bot into why you're doing it, what's your risk appetite, what are the most important success factors and uh, success factors in the short to medium term for you as a business, which then, which then drives which data items are most relevant for you to make sure you're governed and you've got good quality, um, good quality checks, et cetera. So that's the top down, bottom up is the ground realities of what goes on operationally across the organization. Can you actually trace how data flows through the organization between departments, between SILOS and making sure you've got controls around it and being able to manage it and expose it through. In the case of data quality, data quality dashboard. So I think the mix of the top down and the bottom up is important. It's not. You can't just do one, you have to do both. And then overarching all of this, there's the approach around how you govern all of this. And this again goes into do you have the right policies, do you have the right roles and responsibilities? Data owners, data stewards, education and training. Fundamental. Catherine mentioned empowering uh, everybody to do what's needed. Uh, I think there needs to be the cushions and the guardrails of making sure everyone understands what good looks like and the benefits of it. Uh, and then the governance and the committees to make sure all this is looked after. So what I've just explained is your, I would say your typical well governed organization that's treating data quality management and data governance properly. Another side of it, when you look at it from a tech, tech angle because sometimes or historically organizations have sometimes treated this as uh, a technology problem and there's a lot of off the shelf products and tools that companies can invest in. Um, and um, our point of view on it is it works for some organizations but with the extent of AI adoption but also organizations going through their cloud adoption journey, uh you need data quality, uh, your data quality capability to work alongside the ingestion of huge workloads of data which organizations are now using uh, or trying to bring together in order to do uh, you know, some clever use cases with AI. Um, what you tend to see is a lot of the off the shelf packages the organizations have invested in. They have great use for interface but they are not necessarily built for processing power or flexibility. And as you can see the market in the world is moving very quickly and organizations have to pivot and change strategy tact, react to a lot of um, changes in the market quite quickly. And some of these off the shelf products and tools um, are bloated and it makes it incapable of doing so. So you see a lot of organizations, organizations going native and trying to build their own thing or um, and or use tools and scripts that make it easier rather than this big uh, giant technology led transformation programs. I think you see the shift into more business led transformation programs and which is why you know Katherine's team has been usually successful at Cove as well.

Speaker B: Yeah, absolutely. And um, you know Joseph raised something there about the kind of the design and the implementation of data and AI platforms and how that actually impacts the people using them day to day, um, and the ability to make sure that they actually meet the needs of the end users. How have you found that, uh, Catherine, particularly in the insurance sector?

Speaker D: Yeah, thanks, Gareth. And I think this is something that often gets overlooked and it's worth remembering that, um, according to some of the top consultancy firms, between 75 and 90% of digital and data transformations fail. And one of the reasons that uh, is a driver of that is a lack of alignment on those business outcomes. Um, from my perspective, I actually think it's quite a simple answer, which is to set up your program and um, your maintenance of that program to ensure that your end users are involved in each step of the way. I think there's always a bit of a temptation to let your end users give you your business requirements and then let them go about their merry way with the promise to check in with them, um, uh, when you deliver them these fantastic fancy things. But there's a real risk there of a translation failure between the needs of the end user and um, those engineers who have their hands on the keyboard. Um, unless you've got a really big bulletproof, um, requirements and refinement process, um, it's always good to help mitigate that by having your end users involved, um, at each step of the way. And I think creating um, that culture and strong set of collaborative involvement with your end users, um, can really help shore up your delivery and ensure that the asks that um, you've leveraged on the program are the ones that you're actually getting. In the insurance sector. This is particularly important. And um, if I just talk uh, for a moment about the UK motor market, which is one of the most competitive financial markets in the world, good data in the right structures in the hands of your end users is an absolute must have. You can't be successful without it. And um, in addition to that, as a regulated sector, um, we also have a responsibility to demonstrate an evidence that the data and AI products and outcomes that we're reaching are explainable and that they lead to good customer outcomes. And this really requires the implementation of good governance frameworks, security and privacy by design. So, um, there's lots of benefits, I think, to structuring uh, your program or your transformation in a way that involves end users, um, at each step of the way. And I really believe that that gives you a higher chance of being in that 10 to 25% of transformations that succeed. Um, I just don't see a downside for it.

Speaker B: Yes, it's a terrifying statistic that one, isn't it? You feel like you're facing uh, an impossible task. But I thought that um, was really fascinating about bringing people into the process. I was wondering how you went about doing that. Do you choose uh, individuals to kind of work with you on the way or do you have panels of uh, colleagues who you bring in? How did you kind of approach doing that?

Speaker D: Yeah, so clear roles and responsibilities, which Joseph mentioned a little bit earlier, is one of the key aspects that have helped us be successful. And we're very clear internally on who owns the what we're doing and who owns the how. And to be able to leverage those two things together. You've got your experts, your technology people who are responsible for the solution, design and how we do something. Um, but you've got your business and your end users who are setting the what. Um, those two things are equal to each other. There's not uh, um, a kind of fight uh, for superiority there M But they do need to um, have equal weight. Um, so for us being really clear who is the what and who is the how, um is the first step and then underneath that um, we've been engaging at each step of the project with um, the wider stakeholder group so that we avoid that situation that I said where at the end of the program you put into the hands of the people who have asked for something, something you think is fantastic and delivers everything with a bow on top but actually just doesn't cut the mustard. Um, so whilst we have leads who are responsible for that what and that how, there's a whole set of wider stakeholders that are, we know, we report to and um, you know, do show and tells for demos for that, that sort of stuff, um, that allows us to ultimately converge to the right outcome for uh, both the solution and the requirement.

Speaker B: That sounds like a very collaborative process. I'm sure that's really key to making uh, sure that these um, these programs land effectively. Um, you know, another big uh, failure point I think for transformation initiatives is the element of scaling. You know, perhaps you've got a working pilot. Bringing it across the business can be challenging, uh, especially uh, you know, when you're trying to bring it up to full scale deployment across the business. So um, Joseph, working with your clients, uh, what are some of the key barriers that you've observed as companies go through this transition and how can they be overcome?

Speaker C: Uh, thanks Gareth. One thing that I've, or uh, we notice quite a lot and in fact I've come across that in multiple conversations just in the last few months, uh, with some uh, clients we're speaking to is this, can you come and do a review, can you come and do a pilot or a POC for a particular use case? And as we engage in the conversation we will say, well to get the most out of this, I get the three of you are the working group, but we would like to engage the other stakeholders that are impacted by this indirectly and directly and to understand what success looks like for everybody and try and make the most out of it so that everybody sees the benefit and we make this successful. And most of the time we see hesitance to that, um, and friction around, no, no, no, let's not involve others in it because that will grow arms and legs. Let's do this and keep it contained, um, and just do this one thing. The problem with that then becomes, you can imagine what happens in the next phase because as you try to scale it, you're then bringing together all these stakeholders who are not involved in that initial pilot POC whose opinions were not considered. Um, and the buy in you then need, uh, will naturally be more challenging. The other important point is, um, while the approach of let's experiment and make sure we get some benefit before we scale is absolutely the right thing, but it's also being cognizant of the buffers you put into what it involves in scaling a successful poc. No one goes and does a POC on something that's complicated or you don't pick a complicated use case. So one example is from a business perspective, everybody would love to get a single view of the customer. Do we understand the customer? Do we understand all the products and services they are using and how we can do more with them? Um, that's not a use case for a poc, um, in first stage, or you may do a little bit of it, um, you're going to take something that's low hanging fruit, that's easy to prove, where you've got some data and you can quickly show a benefit when it then comes into, let's scale this into the benefits and do something like single customer view. You suddenly realize that's not something you can do in two or three months because it then goes back into what I've said previously. Do you even understand where all your data is? Do you even understand who owns what data in which data system and source in which department so you can actually start pulling it all together to understand your customer? Do you even know what's in it? Do you even know what's the right data to you, so it goes back into the foundational elements and even if you knew the answer to all that, do you have decent enough data quality? Can you trust the data using? Do you then have the infrastructure and everything set up in order to do something with it? Do you have all the right stakeholders bought into this uh, as to what single view of customer means for them and what benefit they get from it. So that's an example from a, from a, from a pure business use case example of why it becomes a huge challenge. Another one is, you know the easiest thing these days with how much cloud and Microsoft and Amazon and Google have advanced and how quickly they're innovating is when it comes to infrastructure. And it uh, a lot of the past hurdles uh, have been, have become much more efficient now if you want to deploy something. However, one common use case we see a lot now is organizations that are on this digital transformation, data transformation journey trying to do more with AI and trying to adopt the cloud and various features is we've got all this data sitting here, we need to move it across. But the storage costs uh, are unbelievable. So we need to now look at what data we've got. Why are we moving it? Do we have personal, do we have confidential information in it? Do we have sensitive information? And it again goes back to the question of do you even understand all the data you own and where it sits and what's in it? Um, so again I'm going back to the same point again. And so I think the summary is it is getting those foundations right. Do you have the right processes? Do you have cultural buy in from the right people, from the top? Does everyone understand the value of why you're doing this? To Catherine's point around constant tangible benefits, do you have the technical foundations in place? I right access to the right data, uh, in the most efficient way. Um, uh, and more importantly do you have the right people and the right talents to make the most with it?

Speaker B: Do you find that um, the barriers to scaling successfully tend to be a mixture of these things or is there one barrier in particular that you see as the most common and the most pressing For a lot of organizations I

Speaker C: think it's a combination of things, but none of, if you have to pick one thing and I think this is less of a factor now, it was a big factor a couple of years back is uh, the leadership or the senior team or the board or exec, whatever the construct is of the organization understanding the importance of it and buying into it. Um, because I would say you can have Everything else you want, but without that you can't really put fuel into some of these initiatives. Now that said, just that on its own doesn't help, but I know that was a big issue in the past. I think right now everybody, it's more the opposite. You probably have more pressure from the top around. What are we doing about AI? What are we doing about data? It's probably the opposite. And then whether the organization's ready to move, which then goes into the foundations.

Speaker B: Yes, indeed. Um, and I think most of our listeners can identify with why that might be and why the um, the general atmosphere around AI transformation has changed in the last couple of years. Um, uh, Catherine, you know one of the things that Joseph mentioned there was buy in, you know, from, from the top and um, and I guess from, from all elements of the business, um, from your perspective, you know, this speaks to the kind of like cultural and operational challenges that companies face as they launch, um, data and AI initiatives as they maintain them. Um, how did you approach these kinds of operational and cultural challenges, uh, to make sure that you overcome any resistance?

Speaker D: Yeah, and um, I think there's a common theme here which is to say that data and AI initiatives have the potential to really touch almost every part of your business. Um, and to Joseph's point, they therefore need real buy in um, from the decision makers. Culturally, I think there are a few challenges to keep in mind. Firstly, I think there is a potential to actually limit your ambitions by just rebuilding what you have or just um, migrating what you have and not really leveraging the power of more modern um, toolkits and techniques. You know that famous analogy of Henry Ford which is, you know, if you ask somebody how to get somewhere faster, they'd say I need a faster horse rather than a car. So I think just um, culturally making sure you're not limiting your ambitions by um, just using your past experience or your past um, uh, infrastructure, uh, can be a key one to overcome. Um, secondly, you need to make sure that there is a training program involved in any transformation. And when you're using data and AI to progress and ah, make your business better, then you need to bring your people with you, invest in their training and the opportunities that that brings. If you don't do that, that can lead to incorrect conclusions or um, uh, you can even get into like sabotage situations where you just, your people aren't with you. Um, and culturally, uh, I think there also is ah, ah, a very high risk of scope creep. There's an assumption that a data and an AI transformation will pick up and solve every single problem that the business has. Um, and you need to be very careful and very strict about where you point those resources and um, what you're delivering in order to keep that momentum operationally. Um, there's a few things to be mindful of as well. My team laugh when I say this, um, but I say we're building a data lake, not a data swamp. So we need to make sure that we're very structured uh, about what we're doing and how we're doing it because it can get away from you very very quickly if you're not careful. Um and part of that can be a lack of delivery in search of perfection. So make sure that um, you are uh, breaking it down into the uh, smaller chunks that you can deliver on a more frequent basis in order to continue that perfection. Um and that can also limit your inability to scale, especially when it comes to AI. Um how we've approached this. Firstly we haven't tried to do this all of ourselves or by ourself because um, as an organization we've never done this at this level before. We've had programs um, in the past but the scale of ambition that we have is high and we've not done it before. So we've partnered with, with experts to help us do this. We're working with Dufresne, we're working with databricks, uh and there's some really fantastic um, ready to use uh, platforms that can help you start that journey and build those foundations. Ah and critically that helps you get it right first time. Um because uh, if you're working with experts in collaboration with your internal team, you can uh, you can really uh, go much faster and learn and leverage those expertise to help um, you help get it right first time. Uh, which I absolutely recommend by the way. Um, and that can also enable you in your future ambitions to scale because you just get the foundations done and dusted in a way that um, is least painful and uh, most successful. Um, with uh, limited challenges. It's not easy. Um, but working with a set of experts can really help there.

Speaker B: Absolutely. Um, before I ask uh, Joseph what his ah, take on that is, I must chase your analogy and ask you what would you define as a data swamp and how does one avoid finding oneself in it?

Speaker D: So um, I, I'm a big, big believer in um, having a simple approach as you possibly can when it comes to the design of your data model. So that's the uh, and I'm going to use another analogy. If you don't mind, um, which is a baking analogy, um, so you, you need to know all of the recipe, all the ingredients and the recipe in order to bake your cake. And a data model and design, um, is the recipe for which you put your data into that ultimately gets you to hopefully a nice cake. If you don't do that, uh, that's where I'm suggesting that you can end up with a data swamp where you've, with the best intentions, you've brought a bunch of data in, you've got ambitions to build AI models or um, other analytical tools off of it, but you've not taken the time to transform or structure that data in a way that makes it most efficient and most easy to use by the people who need it. Um, so, I mean it's quite an extreme analogy to use, but I think it's effective, um, because it helps people realize the value of that design in collaboration with your business users as part of that, um, foundational process, um, and not retrospectively when you suddenly find yourself with a lot of data, but in a shape that isn't actually useful.

Speaker B: Absolutely. I love that. Um, Joseph, do you have any, uh, take on that regards to overcoming those cultural and operational challenges?

Speaker C: Um, I was just going to. The one thing that you've, uh, made Katherine repeat there, and I'm just going to use that example there, is the data, uh, how not to get to a data swamp. One of the things we see across various bits is, um, and this is an operational piece around maintaining what you've built, a great lake. But if you don't maintain it and your focus goes into other bits or the operational efficiencies around how to, you know, have a good lake and continue having it, if those principles are not adhered to, it quickly becomes a swamp. Which is one of the biggest reasons why if you go to any of the data, uh, event circuit in London or anywhere in the world, and you'll hear people come and talk about success stories in their organizations, two years later you will hear another person come in doing the same role in the same organization, saying, everything was a mess and now I've sorted it with this new stuff. It's because things have been done but not maintained. Uh, so I think that's a core element of the operational excellence principles of, again using Henry Ford's example there, that the automobile manufacturers perfected back many, many years ago with Lean six, et cetera. A lot of those principles are still relevant when it comes to bau business as usual, keeping the lights on, running the operation type Activity Just thought I'll rather than talking about um, all aspects of it, Gareth thought I'll use that example and give a uh, specific view.

Speaker B: I thought we were going to get into baking talk for a minute.

Speaker C: Not my territory. Sorry.

Speaker B: Um, well in that case let us um, talk because Katherine um, works for an insurance company in the uk. It's a highly regulated industry and I think ethics uh, around use of uh, data and AI always really important topic in this area. As the technology continues to evolve and move into different areas of the business. How do you think companies should prepare uh to handle these evolving technologies and the ah way they're implemented?

Speaker C: I think the biggest thing when it comes to compliance and ethics and uh, privacy, security, data security etc. If you look at um, as we said previously, boards, companies, everyone's experimenting with AI and how to make most of it, how to adopt and scale it. Um, and if you actually look at the biggest obstacles in the way outside of data foundations which we won't talk about anymore, um, is uh, managing risks and regulation around privacy compliance and ethics as a consequence and as an additional topic there. So it is a huge a topic that's getting a lot of focus uh, currently across organizations. If I first talk about the privacy and the security side I uh, think pretty much all solutions now and anyone embarking upon a transformation initiative or a change initiative, uh, security, privacy, uh, compliance is sort of by design as opposed to something you react to. Um, and that's what we see as best practice when you do a POC or you're delivering an initiative. So I think if organizations are not taking it that seriously it is something that you need to consider in your first step. Now the big organizations are quite mature in this space I would say relatively. But majority of organizations may not have this as a top priority given constrained resources or whatever they have uh, in terms of other challenges. The other side I would say with regards to this is it is quite easy with the tools and uh, technology available today to very quickly do a one off health check to see where you are. We see a lot of organizations reaching out um, or looking for help in this when the auditors raise an issue or the regulators raised an issue around do you have controls around the data you use when it comes to are you complying with um, privacy and security and compliance rules around what data you use. With AI as an example, um, and it doesn't have to be that difficult, it's quite straightforward and simple to do a check. Just like the data quality examples I used previously, you don't have to go and invest in expensive, uh, off the shelf packages and tools, which is what a lot of big organizations do. And it becomes an IT problem. And there are simple scripts to quickly do a health check of where you are now when it comes to ethics. Uh, one of the things we've seen again uh, in the market, the bigger organizations, the most more mature, uh, enterprises have dedicated teams and leaders, uh, who purely and solely look after ethical considerations when it comes to the use of data, AI, uh, and how it's all embedded into every process in the organization. Um, and these leaders have a seat in the data leadership team or even beyond that in other committees. But I would say a strong AI strategy. The context of this conversation does start with customer trust. I'm sure we all have our own conscious and unconscious bias around various brands, towards various brands and what they do with our data. So I think organizations definitely are all putting customer and customer trust at the heart of um, regulation aside. So how transparent are you with your data principles and policies and giving the customer the autonomy to decide what the organization can or can't do with the data and then equally your brand trustworthiness is equally an important factor to consider in it. But I would say without going into too much specifics on ethical frameworks and principles is uh, sometimes I think while regulation can be a pain, ah, sometimes it also lays out the uh, roadmap or the playbook to follow for those who are confused as to how to start and how to go about it. So in the financial services context, things around consumer duty and from a compliance perspective, things around GDPR and um, what the ICO have come out with. But more importantly and more recently the EU's AI act, uh, which aims to regulate the ethical use of AI and data, I think is a quite a good, useful guidebook, uh, and a path to follow even if you're not, uh, even if you don't have to hit a compliance timeline very quickly. But it's a good guidebook on how and where to start.

Speaker B: Start, absolutely. Thank you. And Katherine, how about you? How do you kind of approach, um, you know, your ethical responsibilities for during these kinds of transformations and more broadly in your data and analytics practice with

Speaker D: um, with great care, um, because there is, uh, you know, with, with the innovations like we're seeing, not just in the insurance industry but more broadly and uh, you know, we can think about that example where another company has not taken the care that they should have, um, and that's led to poor customer outcomes and reputational damage. As well. Um, so the important thing uh, for us is to make sure that we have that good governance in place that we can explain in plain English what we're doing and that uh, the outcomes of the decisions, choices, analysis, models, whatever it may be, um, are, are within that framework. Um, there's a visibility to that and ultimately that leads to the right outcome for the customer. Um, so it's something that uh, we take great care on and uh, shouldn't be overlooked I think for companies who are thinking about starting this journey. Um, yeah, make sure you do have a focus on this.

Speaker B: That's great, thank you so much. Uh, let me just move to a bit more of a forward looking uh, perspective Joseph. Um, when you look into the future, obviously there's been a lot of changes over the last couple of years. I think organizations are starting to really um, scale AI initiatives, starting to see value in many cases. Um, how do you see the future of data and AI and how it impacts business models not only in financial services but across the various sectors that you work within.

Speaker C: Yeah, if I um, specifically hone in on the AI part of your question there. For example. Now if we fast forward a few years and I'm still trying to understand in my head how this may work. But when I listen to some of the so called experts and thought leaders in this space, I was fortunate enough to be in a particular, a private session that was run by someone from Silicon Valley who used to work at OpenAI when ChatGPT came out and they were uh, an investor, um, involved in multiple startups in this space in Silicon Valley and was educating us on the pace of change going on in certain parts of the world, California ah, being one and um, it was sort of dramatic in terms of what the person was suggesting, how we should look at our businesses. I think it works great for smaller businesses and uh, pretty much talking about. But AI should be at the core of uh, what your business is in the future and you have to redesign your entire operating model around that. Not try to fit AI into your current operating model but think of AI at the core and then see what that operating model would be. Because uh, then they were giving examples of certain sectors in certain parts of the world where it's just completely disrupted in terms of people, process, organizational hierarchies, what you need, how you uh, deploy the resources you've got. No, so I think that's, I'm still struggling to fully understand what that actually means to all the organizations we see and work with and understand. Uh, but that's I think in the not too distant um, not too distant future someone once gave a definition of AI which I felt was very relevant. In one way it was AI is about uh, reducing the cost of prediction. I've had a big issue with what's the definition of AI because AI has been loosely uh thrown as branding and marketing as a marketing ploy for a lot of companies out there, product companies, software companies, where even screen scraping tools were called AI tools like five, six years ago. Right. Which I don't uh, not necessarily agree with but if you look at the biggest use cases where companies have in the last many years and even currently right now, if you look at the latest surveys and reports, seeing the biggest benefits with uh, AI and gen AI et cetera is around customer service and customer operations and chatbots and all the website bots and call center uh, uh operations. It's all around there where you see the biggest use cases. I wouldn't say it's just that there are every function, every industry have got successful use cases in this space but the most prominent surround customer servicing and customer experience. Uh, and then we look at what's the biggest benefit organizations are getting from AI right now and in the immediate to near future. It's all efficiencies and productivity gains. There is a lot going on around product innovation and just general innovation. But I think while there are some benefits there, it's not to the extent of what's being gained by just being more efficient. Um again for most cases there it's the ability to get quick access to information. If you think about what most of us may use some of the genai tools at the moment is to get quick access to information. Uh so therefore efficiency, the um from an organizational readiness point of view for this, if you look at tech infrastructure with what's available with the cloud platforms, I think quite mature, uh tech stacks and infrastructure stacks are available. But when you look at it from a data management perspective around quality, again the same points we discussed previously, majority of the organizations don't think they're fully ready to make the most of it. And then when you look at who then has the talent or has the pipeline to generate the right talent or develop the right talent to make the most of it, the proportion of organizations that are ready to make the most of it at scale even reduces um, quite further. So you see a divide between companies that are fast moving ahead and I would say everybody else um, in some of this and then the other topic we touched upon, security, privacy, data quality have become huge Priorities for all organizations. And you can see uh, it's top of the list investment area for majority of the organization for this. But if I was to give um, one example which I found, in fact just this month I heard about this is one of the largest banks in the world, not a British bank. Uh, the head of data engineering there was talking about a use case they did while they were trialing various things with Gen AI and AI and it was for the uh, quarterly investor presentations and um, shareholder meetings where they were announcing the quarterly results and they inputted all the Q and A and board results and quarterly results from the last 10 years into the this UM LLM and try to predict what the questions are going to be at the next quarter. And uh, with the latest release I think of it was the ChatGPT, latest turbo version, uh they actually got 100% accuracy of the Q and A, the questions and the answers. Uh, this was two weeks back. Um, so you can imagine what's happening in that organization now. The team involved in it. I've got funding to do whatever they want and huge tangible benefit all of a sudden. Let's try the next thing. The PR departments, uh, the public, um, PR and comms department are talking about um, lots of investment in how they can do more with this. So that's a siloed example there I've given but I think shows the power of what's possible.

Speaker B: Yes, fascinating to see what can happen when those tangible benefits are really demonstrated. Um, one of the things that Joseph mentioned there Catherine, was about how in these, I don't know if we can still say we're in the early stages but a lot of the initial gains um, of data and AI transformations are kind of often internal or um, about efficiency, productivity within the organization. Is that something that you see in the insurance industry and how do you see the future of data and AI in the insurance industry?

Speaker D: Yeah, so I'd actually say it might, might be a little bit the opposite way around in the insurance industry. I think insurers are only just really um, cottoning on to the efficiency piece, especially with the advent of Gen AI, um to help really on a more operational basis, um, to drive that level of efficiency. But um, in many insurers analytics has led the way um for the use of machine learning models and now the appetite for more complex, complex AI approaches in order to be able to compete um, from an underwriting and a pricing perspective. Um, now that's not to say that the insurance industry is leading the charge here. We're not but um, actually I think the scale of opportunity um for insurers is now twofold. Um, how do we continue to drive those gains um, through the predictive power and analytical um, tools and techniques that we employ whilst also investing and rolling out on um, more of those efficiency gains. Be that in a call center or um, in things like chatbots and things like that, um, as well, um, what I would say from a coveyer perspective, um, one of the things that we are highly known for is our great customer service. Um and we do take a personalized approach and um, touch to that. So as an industry maybe there's a move to efficient um, uh efficiency, uh initiatives. Um however we still see the benefit and approach of uh, being able to call somebody up on the phone and talk about your case or the underwriting processes that we go through.

Speaker B: Absolutely. Um, well the time is absolutely flying by. What we like to do, just as we come towards the end of the conversation is to provide some advice and some um, practical steps that people in our audience can take when thinking about these kinds of transformations. So let me come back to you Catherine. What if you think about giving some advice to our listeners, what would you recommend as being the first practical steps for business leaders who want to leverage data and AI to drive business value without getting lost in the noise of new technologies and trends?

Speaker D: Yeah, absolutely. And there's three things that I would suggest. Firstly, be really clear about your vision, write it down and then use that to get uh, the support and uh involvement of your senior stakeholders, um, to ultimately kick off any type of uh, program that you want to launch. Secondly, um, I would highly advise partnering with experts. Um, if you've not done this before, there are plenty of options out there and they really are accelerators that will help you get it right first time and that can open the door for future investment. Um, unless you're really experienced in this, don't try and do it all yourself um first time round, get some help and get some support. Um, and then thirdly, um, I think you need to familiarize yourself with the non technology needs. So not forgetting the internal processes and frameworks that are needed, needed to leverage the power of data and AI in order to be successful and to really um, deliver the value and benefit that you set out to.

Speaker B: That's great, thank you. And definitely practical advice. Ah, coming to you Joseph. For leaders navigating their companies towards uh, greater data and AI maturity, what advice would you have have uh, in order for them to facilitate that transition effectively?

Speaker C: Yeah, first and foremost is understanding the Business strategy and the business goals, uh, because whatever you're trying to achieve has to be linked to it. And uh, you have the it has to help with the business strategy and the business objectives. And aligned with that is ensuring the people at the top of the organization, the leadership, the execs, the board, they not only want to do more with AI, but they actually understand what it entails and um, the time it will take to see the benefits. I think that's number one. Number two is similar to Catherine's point around accelerators and yes, big bangs. Let's try and do uh, a two year transformation is not what the market's doing, not what works at the moment. There's more effective ways to prove value is experiment, do those proof of concepts, proof of values, try and show tangible benefits quickly, even if it's in small pockets. Uh, and again ultimately without demonstrating business value linked to your strategic objectives, uh, most of these initiatives won't succeed. And then uh, finally don't forget, uh, the foundations of what is needed and that is across various aspects of people and organizational culture, uh, etc. But if I just talk about data, if you go back to 2018, 2018, 2019, everyone wanted to do data science and be a data science company. Right now everyone wants to talk about AI and be an AI company. One common theme that's prevailed throughout the last 10 years is get your foundations right around data. So uh, that should always be there uh, alongside everything else.

Speaker B: That's great. Thank you so much, Joseph, Katherine, your insights today have been incredibly enlightening. Thank you both for, for sharing your expertise around data and AI transformations with us, um, to our listeners. I hope you found this discussion as stimulating as we did. Uh, and Joseph, Catherine, any final thoughts before we wrap up today?

Speaker C: Sometimes it's good to pause and reflect on things and this chat really helped me to reflect and to hear um, a my own thoughts in my head. But I equally listened to Catherine, um, as it's not often uh, you hear from business leaders talk about data journeys and data with as much authority as Katherine have today. So really enjoy the chat and definitely invaluable.

Speaker D: Yeah, thanks for having having me on the podcast. It's been fantastic uh, to hear the topics that we've covered. Ah, and I hope the key takeaways for listeners out there is that there is definitely a place to start when you're embarking on this journey. Um, and I hope, hope what we've been able to share today can uh, begin um, uh, helping others to really kickstart that and, uh, enjoy it as much as the people, uh, who work in the industry and get to do these things as well. Um, thank you, thank you again, Joseph

Speaker B: and Catherine, and thank you to everyone who tuned in. Be sure to join us next time on the Business of Data podcast for more discussions about shaping the future of data and analytics.

Speaker A: We hope you enjoyed the episode. Be sure to subscribe to the Business of Data podcast wherever you're currently listening. And keep up with us on socials. And for more content, visit us on, um, businessofdata.com be well and thank you for listening.