Priorities of the group data office at Lufthansa – with Xavier Lagardere

If you're not able to explore data minimally, access data, access quality data, you're not able to initiate use cases or produce insights. Hello and welcome to the new episode of the Data Culture Podcast. I'm Karsten Banger and today I will talk to Savi Lagerdjeer, the group data officer of Lufthansa. This is a newly created function at Lufthansa and we will learn about the scope of the work of this core team of data people that are in a core of concentric circles and at the outside we have about a thousand people at Lufthansa in the data and analytics community. We will also talk about the biggest successes but also challenges that Savi Lagerdjeer is facing in trying to support the business functions better with data. We will also inside node learn why he thinks that data culture is the wrong word for the right content so he fully supports the ideas of having to create and having to influence mindset and behavior to communicate around data and data and analytics successes and also encourage and really support data literacy and training efforts. But he also thinks it's the wrong word, at least in the Lufthansa context people would think that he's going crazy. Enjoy the episode. Hello Savi Lagerdjeer. Hi Karsten. Nice to be here. Very nice that you are with us today. I really look forward to our discussion here. And let's get started right away. You are the CDO of Lufthansa and maybe you can tell us a little bit how you came into this position and when in your career you actually started to like data and make it more or less the center of your activities. Right, so indeed I'm the Chief Data Officer of the Lufthansa Group which is actually a multi company group and I've been ten years with the company leading roles mainly in the commercial function prior to my data role now. I started liking data much before actually being employed by the Lufthansa Group as a commercial leader. A background is in software engineering and I worked within the IT world prior to joining the airline world and my first role within the Lufthansa Group was to lead digital commerce versus airlines which is our airline in Belgium and then on led global retail what we call distribution in an online world which is deeply involved with data and how we best position our offers on the right channel at the right time to the right customer and from there as we were establishing the role year and a half ago and really as the as the group was preparing itself for changing world the world where you need to drive and lead with insights I came across this opportunity to basically deploy my knowledge and my experience of the airline business in even wider scope and I'd be happy to talk to you about how wide that scope is now that this pan of activities entails all the airlines but also the cargo and activities the aircraft maintenance activities and the Lufthansa Group. Sounds super interesting we definitely do that before we do that I always ask one question to get to know our guests a little bit better is so what are you enthusiastic about next to data if it's not data? Well I'm enthusiastic about traveling and I made maybe not coming as a surprise working for for Lufthansa and the Lufthansa Group and it's it is indeed true bit far or close there's always a realm of experiences and enrichment that I personally derived from from traveling in and for business so I'm going to Moro to Geneva for a meeting with Ayata among airline peers talking about data actually in a newly formed data task force and I'm really looking forward to this and and figuring not only what we can do as an industry to drive forward but also the personal part and the enjoyment of just taking things from a slightly different perspective and a slightly different location. Excellent sounds super interesting obviously and maybe we can hear a little bit more about that so you already mentioned the scope of your activities let me let me know more about that and maybe you can also comment on the business strategy of Lufthansa you just said you are in a transformation maybe towards a more inside driven business how does that look like for a large company like Lufthansa? Well first of all I would say there is a context in that context is even more emphasized by the by the pandemic that we are going through or we went through if I can say or that we are still going through plus the overall uncertainty that the world lives in so we are in for a vuka world to use the non-acronym of volatile and certain etc and so the general context the general idea for us as an airline group is to make sense out of this uncertain and volatile and ambiguous context in a sustainable way. Contrary to how you would run an airline historically which is really heavily planning related and long-term stable operations we now need to react much faster to changing passenger needs but also to simply regulatory changes or you know country might be open or closed or as we lived it through the pandemic health regulations require us to inform or even take care of our passengers differently from from one day to the next or sometimes you know maybe one week to the next so that that created a context where we need to be able to be not only much more reactive that's a topic in itself in shortening decision cycles and and driving results faster through to the customer so that the customer ultimately can experience a tangible output much faster but also proactive and and basically makes sense not only of the past status but also derive a trend where we can and you know there's probably not one right prediction especially in in the two days where that are very briefly described but at least we need to systematically much more than before peer capabilities that are enabling us to do things like scenario planning or forecasting or prediction to a certain extent and that acknowledgement led to the to the creation of my role first of all to help the group drive towards more data usage one thing we do have and that anyway are not starting from from nowhere we have a ton of data we have data generated and stored everyday of course in always a legally compliant way especially when it comes to personal data handling but even with all the legal compliance that we always ensure we do generate a ton of data about our operations, peer to the flows or the aircraft themselves about our customers we also have the ton of financial data where we still have an opportunity is in using this data using this data systematically in order to create first of all a mindset of performance orientation of insights and conclusions that are derived from facts and and like I already mentioned as well a mindset of forecasting of trying to predict what's going to happen in the next maybe three three four weeks six months rather than assuming that the past will repeat as you may have repeated that was true for years in the airline world with a lot of seasonal logics that would repeal the apply for more yet to another and chances are that this world is behind us now yeah so the yeah increasing changes and agility that's needed and and that actually it's interesting to hear that that's actually led to the creation of your your role and obviously always in this group scenario how far is your reach or how are you organized now is it that you have data officers in each of the divisions of luftanzer or how can you now be effective in fulfilling this vision or this goal we're in the process of of rolling that out to be transparent so from an organizational standpoint it's it is a working progress the setup of our group is that we we actually have I would argue large business units so we're not conglomerate with a thousand and independent entities where the corporate binding function would add you know really all the all the corporate logic we hear rather talking about luftanzer luftanzer cargo very large companies themselves which operate many a corporate functions not completely independently but with a certain degree of autonomy and so there is a first of all a there are two dimensions try and illustrate that one dimension is what can be what can be the added value of the of the binding logic at the group level and how can we collectively benefit from from taking a systematic approach about topics like data sharing which is internal and external so externally you know being sometimes more cautious about how we operate data sharing points with our partners with our providers and sometimes actually more open for the mutual benefit of our partners in our our joint customers when it comes for instance to helping our customers at the airport where everyone benefits from a more open data culture which then helps basically plan the operations and take care of our customers better so there is one layer I was talking to you about two dimensions. There's one layer of let's find out the added value of this group logic in the specific areas of data source automation. And we can come back to that a bit later what these areas are. And then there's another dimension which is what are the roles that are going to be operated from within a company or within a data domain? And what are those roles that need to be, let's say, sitting at a group level? And right now we are in the process of doing the two things, which is on the one hand side structuring our cooperates or group at value adding a layer to make sure that there is commonality, harmonization and efficiency sometimes, but as well group impact in the data activity, as well as we are rolling out the logic of things like data, data, user data producers, data consumers internally with the largest companies within the group. That actually doesn't always take company by company logic. You see, for instance, if I take the airlines of the passenger airlines of the group, so the likes of Austrian airlines, Swiss, Buffetanza, Brussels where I worked, the logic that we follow in order to apply data governance, let's say, in our operations field is likely more going to be following functional areas, rather than going on an airline per airline basis. And so right now, for instance, we're working with the colleagues that manage our ground operations to figure, okay, what data do they consume and produce, not only are they at a local airline level, but across the entire airlines that we can then, you know, document, make available for inside generation and forecasting across the entire company. Okay, so it's a functional approach. So I understood that you are encouraging the use of data, that you are trying to move it forward, but it sounds like you will also try to set standards and see that it's a repetitive task, that's something like ground operations, and the data use in ground operations can then be repeated across all airlines that are in your group. Is that correct, understanding? Yes, it's a correct understanding on the two fronts. So yes, it's a functional approach. And it's also an approach where we are trying to, you know, do things once for all. And, you know, let's keep this example where we are working on the ground operation side as one of our, you know, pioneering functions. Once things are being achieved there, we will take this blueprint and roll it out further. Okay, got it. So what else is in scope? What else do you do in your group data office? Well, there are basically three specific activities that, you know, where we figured, okay, that's adding value overall. One is actually what we just talked about, it's about setting standards and processes, which help functions, businesses, produce good quality data and access good quality data. And the quality is a very important word in there. Sometimes data are incomplete. I mean, always data are incomplete, but they need to be incomplete only to a certain degree. And that's what we're gradually making sure about that we become aware and can qualify what data is ripe for use and what, you know, maybe we shouldn't be using because it's too incomplete. That's all the way to really having a clear role for data owner, data's towards that can share part data, help people who need these data internally, then access the data. So that's one area. The second area has to do with accelerating delivery of insights. And so one of the things I am I'm sure about to be able to have a small but really excellent team of data experts, mostly data scientists, but you know, some data engineers and data analysts in order to be able to basically accelerate projects and drive immediate impact where it's needed. This is not meant as a nano and unit that will carry on projects all the way to production, but really as a initiator sometimes or as a compliment of things that functions or units were not able to achieve by the own or didn't think about on the own. So there's of course the two things go ahead in hand because if you're if you're not able to explore data minimally, access data, access quality data, you're not able to initiate use cases or produce insights. So that's one direction. On the other side, this team that you know, accesses the real data and produces insights, experiences, the limits of our today's data ecosystem. And they are basically the first ones to feedback. Okay, this should be enhanced. We have a problem here or there. So really the two things you have to think about it as a dynamic feedback loop that feeds one another. The third area that's that's actually an affinity here intentionally. It's really last but not least. It's the most important thing. It's I would call it transformation, but it's really about mindset and behavior change as well as trainings. You could also call this data literacy. You know, if you were looking for an embryo name. So that's the third thing at the group level and I can come back in more details. Sure. We would call it data culture and data literacy being one part of it. But also, as you mentioned, mindset and behavior. It's even harder to influence. But in the end, that's what culture is. It's basically mindset's behavior. How do you do things? How do you approach things, et cetera? We definitely have to drill into that. Before I do that, just maybe a quick word, the team you mentioned. How big is that? Well, it depends how you look at it. So I don't necessarily look at who are the people that are with me in a group corporate units. But I'd rather look at concentric circles. So what I call the data community of the of the lift and the group. In its wider sense, we're talking about without getting a precise number, but we're talking about the range of the thousand people that are, let's say, data professionals in the sense of business intelligence data analyst across the entire company or across the entire companies of the lift and the group. So that's the wider circle. Let's say in the data community. Then you would say there's a second or more inner circle. And those are the people that are really developing models that are working on forecasting. And you find these in many areas in the lift and the group that are already data intense. So that's the case in network management, in revenue management, areas which for years have been even leading the industry in their field. And we're probably talking about maybe a hundred or so colleagues across the entire group. And then you have this transformation team at the center, which we were talking about, you know, what you could call my change team. That's like we said, trying to put it all together. If we were talking about a very modest team, it's, you know, 20 or so people. Could you give me a little bit more detail on this core data team? So what competencies did you actually assemble here to bring data forward at lift answer? Well, obviously it's it's it's a diverse array of competencies because you can't, you know, have only one scale. But if I would focus on a couple of of profiles, one are colleagues who are able to think in systematic ways, process, orientations, standards, or rotations. So I have people who are looking at means to simplify, harmonize, and, you know, then explain as well, coach change. So in the central team, that's on the one hand side. And then on the other hand side, I've paid attention to, like I briefly said, to hire top-notch data scientists because I really wanted to make sure that we would be able to drive a contribution towards the business use cases that would be expressed if the functions or business do not have the capabilities that, you know, it's possible to complement the team and prove the value right there. So you many people inside the company see this because that's how they experience the day-to-day of my team. You know, it's it's the team of basically collaborative individuals that join them in projects as the as basic to solve difficult challenges. Yet, I think it's important to me that I mentioned to you the the part I started with. So it's not only those profiles, but also really people that are I actually find it's quite hard to specify the profile to be completely honest. I was recently recruited for this team and then we changed the the job at a couple times. Now we call the data excellence and we want to but then it's you know, it's really through the discussions with the people that we are meeting for interviews that we can really figure if the person would fit our notes. It's rather about what certain behaviors, what to how you think, you know, if you're going to think in systematic replicable processes and that sort of things, then that that'll fit. Yeah. So okay, you mentioned that you have people that are supporting the setting of standards that are coaching that are thinking about processes. Then you have the data scientists that are accelerating projects contributing to business cases or use cases. What about the third area you mentioned as your scope? That is mindset, behavior, data culture, data literacy. How do you take care of that? Well there, first of all, I wanted to come back to what you were saying about data culture. I found it really interesting that you obviously picked up from what I was saying that I didn't mention data culture. I'm not I'm not sure about the word because I think it could mean many things to different people and I was intentionally and I still am when I communicate internally. I'm trying to be more specific that this sort of umbrella culture world, although I totally relate to how you how you call it. But then if I use if I carry on on this to break down the task, there are different different things that I'm looking at in there. First of all, it there is classical training design and they're actually I could actually use more people. So I'm still on top of that. until recruiting with two main purposes. That's basically increased the general capabilities on the one-hand side, or let's say at the floor, and then increase the top capabilities to always be market cutting edge, and have people that are able to work with the latest trends, latest technology, and that would be sort of increasing the ceiling, pushing the ceiling. So our training programs, which we are, by the way, not finalized with, so we are, right now, rather a step of, we know what personas we want to address with training, and we are in the process of designing the syllabus for each of our personas that we target. But basically, that you could argue, these are tasks that are common to many fields, data is just one of them. And here, I want to accelerate, and I have people that are able to sort of project manage these training efforts. And then, there's the whole topic of, let me use your word, Culture Change, and then it's about, how do you influence adoption through a variety of means. I'm a strong believer that an example speaks a thousand words, and so in a way, one large part of that mindset behavioral change is going to be about making sure that we have a good funnel to showcase the results of the practical work that's being done elsewhere. So whenever there's a good, let me give you a complete example. So we managed, for instance, to improve the way we foresee offloads of our cargo capacity. That can happen, and if you follow the airline industry, you know that as a business logistics cargo has taken a very important share of our business during the pandemic. We carry a lot of goods everywhere in the world, and that's really become mission critical as well, in many aspects. And many times we need to take off certain parcels and replace those with others. And so it's become very important to be able to better forecast this. And we've been able to do this, we've been able to deliver a model that's actually proving results on bench by comparing with actual facts. And so basically, what I'm doing is actually what I'm literally doing with you now. It's showcased that story and show how it brings value. That's, in essence, what I at least personally truly believe is driving mindset change. - Yeah, completely agree. In our framework, we call that data communication. It's really talking about it, not only educating people, but also talking about successes and really time and time again, bring it to the attention that data is valuable, that you can do great things with it. And very important that it supports the business goals. Where it really helps to achieve strategic business goals. So I completely agree to that. And I think it's a good idea to really highlight that. Let's take a little bit of step back and talk about maybe the good and the bad. I would like to know, first of all, maybe the good start with the good. If you look at across Lufthansa and your achievements today and what you're still planning to do, so what are the data capabilities that you have developed already? Where do you think you're really leading and where is still room for improvement? - So I'll follow you taking a step back. And when I look at successes, I'm trying to think, okay, why are replicable capabilities? There's a couple of fields where we are making a difference. For instance, in the area of machine vision, computer vision, where we have really a good technology now, deployed to watch, in our case, aircrafts, when they are in the TAMAC, and monitor everything that's happening there, take time stamps. So we're in the process of talking with our partners at our different partner airports to see how we can roll that out further. But from a pure thinking point, that piece is fairly mature. And I would argue it has potential to be deployed even in a broader context. And if you can monitor an aircraft, you can probably monitor a warehouse or a logistic chain or a production system. So there we have in-house developed a pretty good basis of technology. And the other area where there, I would say we are using capabilities in a way that is bringing value is NLP. Actually coming from the shop floor of our booth and the technique subsidiaries, aircraft maintenance, it's actually the world-wide leader in aircraft maintenance and repair. And so there is a strong basis of natural language processing to interpret inspection results, which we take further, which we can daily take further in passenger-facing use cases, for instance, can we potentially better take into account the feedback that the flight attendants are giving about a certain flight. And that's often done in written formats, natural language format. And then we are currently looking at how we can use this module in order to categorize the results. And we think we can drive really concrete outcomes there. I'm a different confident. But you're also asking about where we have room for improvement. And you see actually, I feel confident that once we have good data and when we have good people and we do have good people, we actually can lead with applying this data. So where we struggle and we can still improve is actually uncovering good data. So that actually speaks about rather more basic things like data access, like making sure that data is labeled properly or reaches the consumer with what I was talking about before the proper quality. I don't have group of all numbers, but, for instance, out of one of the main data warehouses that we're using for commercial purposes, we are on a daily basis, we're only using 16%, that's one 16% of the data that's getting stored into this data warehouse. So that means if you just take these daily snapshots, we have a 84% under utilization of the raw data putting in there. So there I see, of course, a massive room for improvement and that goes through simplifying access, making sure that once you want to enroll to the entire environment, it's a matter of hours and not a matter of days. And then when you access the data, you can work with it. So that's also initiatives like to be concrete data catalog, for instance. We're looking at that, how we can, I was briefly talking before about question, how we take a functional approach to deploying users roles, and we're basically doing the same in parallel to deploy data documentation in our catalog. So as we go and say, let's have systematic roles so that you can all and produce data better in parallel, we will say, okay, and by the way, let's make sure that we document your data. - Yeah, that's, I mean, that's a process where a lot of companies are in right now. We actually make this the main topic of our newest data culture survey. We're looking at data access and including data cataloging. And did you employ specific ideas or methods to make this data catalog really usable, especially also for business users? Did you spend some time in looking at that because we do see a big concern that a lot of data catalogs in the end turn out too technical. It's more like a database field catalog for the developers or so. But did you look at that topic? - I could totally relate to the challenge that you're describing. So without pointing in one of the other solutions, but I would say our answer to it is to address it from the business producer of the data. So we do not do this by talking to, let's say, just a data engineer, but we're making sure that at least the data's towards, which in our case is most probably sitting on the business side, some exceptions, is leading that labeling effort. - Yeah, okay, yeah, makes sense. And was it easy for you to identify and find those data stewards? Did you have to incentivize them? I know that some companies really struggle in that that often when data is not really the primary concern of business departments, was that something that was easy or did you find it also quite hard? - Look, I would take a little bit of a step back and I would have three things to say through this. First of all, rationally, nobody debates that this is a crucial area of success, right? So the minds, my second point is what's missing probably is still to get emotional about this topic. I didn't win the hearts, so too sick to speak even. Without talking about specific incentives or mechanisms to recruit and enroll and maintain the roles of data stewards, broadly speaking, I think I, in the first place in the media as a company, we still have to, so to say win the hearts and make sure that we see and collectively we see that the topic of data is the topic to even shine with as a company or as an individual. And third, I just wanted to be sort of transparent and realistic about the stage we're in. Like I told you before, we are the stage we are pioneering the approach with a few targeting domain, targeted domains, who are sort of volunteering themselves, right, saying yeah, and for me, data is actually crucial and I want to work with you in order to make it better [BLANK_AUDIO] I domain and the stakeholders that I'm working with. So I'm afraid that I have not yet hit the point, which I will certainly, where it will be difficult to recruit because I'm reaching those domains which do not see that crucial. - Yeah, okay, well, thanks for sharing these insights. Makes a lot of sense. All right, one last topic I would like to ask you again. You, in the beginning, you mentioned, I think you mentioned that you started to build a group data layer. And then now this is always a topic of discussion. And especially in group data offices. What about the data? Do you basically create maybe another a central source of data? Do you create a new data platform? Is it more data fabric? I really try to integrate what's already there? Can you share some thoughts around that or what you're doing there? - Yes, sure. And actually, I can maybe clarify because I'm not building a group data layer. Like I told you, also, we have rather large business units across the group. And so far, we had a varied landscape where various solutions, various cloud scalers are being used for different purposes. The obvious question is, would it be better to have just one? Yes, in an ideal world. Did we start from scratch? No, so what's the focus? Well, the focus is actually to use the data that is there. Of course, to bring things to the cloud, which itself bring efficiency, more security, so all the cloudification agenda. But not necessarily in a one-size-fits-all logic, which wouldn't even match the reality. So more important than this. And if I can really emphasize one, to make sure that the data that is being stored, wherever it is stored, is made accessible, can be shared across applications, and can be used for business purposes because it is a good quality and well-documented. Yeah, okay, great. So let's finish this off. And thanks really for your insights and giving us this great overview on what you're up to at Lufthansa. Maybe let's sum it up. I mean, you made clear where you are coming from as a business, which is I think you said really data aware and data does play a role, and you said, nobody debates it, which is good. We do know a lot of business where it's still a bit of a debate as a data should actually play a big role. And then you basically explained the scope. Where you're working, setting standards, accelerating projects, showing concrete value, and then working on the, I call it still culture, so mindset behavior literacy. And that showed basically the scope. And you also mentioned the capabilities that you have, I found it really interesting, machine version and NLP. Not a lot of companies are really saying from themselves that they're leading an NLP, which cause it's sometimes quite a difficult topic, but it's a lot of sense in the way you explained it. And also in your challenges, which are governance data quality you mentioned, but first and foremost, you're simplifying data access, looking at data cataloging to also bring the data easier to the business users and similar initiatives. Anything that we missed, anything that I missed in the summary, anything else that you think should be mentioned here. - Well, first of all, thanks for the exchange question. And that was a great recap. I think you picked all the points in your summary. I just maybe would like to add that across the Loftanza group, we do have a wide variety of profiles and colleagues that are contributing to these activities. And you know, that we are, I think it can pride myself for having very interesting challenges to address that are delivering tangible value to across really a lot of stakeholders and people. And always welcome people who want to help us with those challenges to come talk to us and apply to one or the other job postings. - Absolutely, everyone is looking for talent. But let me say, really good luck with your efforts here. It really was clear that you are more at the beginning of building this group data office. And with that good luck, and maybe at some point in the future, we can hear back from you how things develop. Yeah, with that, thanks a lot for sharing all this insight with us and good luck to you and bye-bye. - Thank you, thank you for having me. Bye-bye. (upbeat music)